# The Internet Ninja

> An independent verified-proof platform for AI and process-automation outcomes. Every case file states its sources and verification method on the page; figures are quoted from cited sources, checked against the live source, and archived. Nothing is invented or rounded.

Every case file carries one badge, and only the first means verified. **verified** (green): independently checked by The Internet Ninja against the public record, at confidence 0.95 or higher on every critical claim, with the method on the page. **corroborated** (teal): closed with real independent sourcing, but the record does not reach the verification bar. **unverifiable** (amber): closed because the public record is too thin to confirm or refute the claim. **pending** (amber): still being checked; not a verdict. **no public proof** (red): the claim circulates, and the public record holds no proof of it. A badge never depends on the subject confirming anything: The Internet Ninja does not ask a company to sign off on its own numbers, because that produces a testimonial rather than an audit. Payment, placement, and relationships never affect a badge.

Every link below points at a page's markdown twin: the same URL with a .md extension, front matter (title, description, url, updated) and then the page as clean markdown. The HTML page at the url in the front matter is canonical; cite that.

## Verified case files (183)

- [AI chatbots and earnings: Denmark study of 25,000 workers finds about 3% time savings and no earnings effect](https://theinternetninja.com/stories/denmark-llm-chatbots-adopted-across-11-exposed-occupations-measured-time-savings.md): AI chatbots and earnings: a Denmark study of 25,000 workers finds about 3% self-reported time savings and no effect on earnings or hours.
- [Da Silva Moore v. Publicis Groupe: the first federal judicial opinion to approve predictive coding (computer-assisted review) for e-discovery (S.D.N.Y. 2012)](https://theinternetninja.com/stories/da-silva-moore-v-publicis-groupe-sdny-peck-first-judicial-approval-predictive-coding-2012.md): Da Silva Moore v. Publicis Groupe (S.D.N.Y. 2012): Judge Andrew Peck's opinion, first federal approval of predictive coding (computer-assisted review) for e-discovery.
- [AI found a new antibiotic: an MIT deep-learning model screened over 107 million molecules and discovered halicin](https://theinternetninja.com/stories/halicin-mit-deep-learning-antibiotic-discovery-107-million-molecules-cell-2020.md): MIT (Cell 2020): a neural network screened over 107 million molecules and found halicin, a structurally novel antibiotic that killed resistant pathogens in mice.
- [ICO: Royal Free NHS Trust broke data-protection law sharing 1.6 million patient records with Google DeepMind for the Streams kidney-injury app](https://theinternetninja.com/stories/royal-free-deepmind-streams-ico-1-6m-nhs-records-data-protection-2017.md): In 2017 the UK ICO ruled Royal Free NHS Trust broke the Data Protection Act by sharing 1.6 million patient records with Google DeepMind for the Streams app.
- [Kadrey v. Meta: a court called LLM training on pirated books fair use, but only because the authors failed to prove market dilution](https://theinternetninja.com/stories/kadrey-v-meta-nd-cal-llama-ai-training-fair-use-market-dilution-summary-judgment-2025.md): Kadrey v. Meta: a court called Llama training on pirated books fair use, but only because the authors failed to prove market dilution.
- [A federal court held judicial immunity applies even if a judge 'relied wholly on artificial intelligence' to issue a ruling: Phillips v. Parlade, D. Nev., Aug. 12, 2026](https://theinternetninja.com/stories/phillips-v-parlade-d-nev-judicial-immunity-ai-wholly-relied-ruling-dismissed-2026.md): Phillips v. Parlade (D. Nev., 2026): a court held judicial immunity applies even if a judge 'relied wholly on artificial intelligence,' then dismissed with prejudice.
- [An MIT randomized trial in Science: ChatGPT cut professional writing time 40% and raised graded quality 18%](https://theinternetninja.com/stories/noy-zhang-mit-chatgpt-writing-rct-40pct-faster-18pct-quality-science-2023.md): MIT RCT in Science: ChatGPT cut writing time 40% and raised graded quality 18% across 453 professionals, narrowing the gap between weaker and stronger writers.
- [GitHub Copilot in the field: three company RCTs, 4,867 developers, 26% more completed tasks](https://theinternetninja.com/stories/generative-ai-copilot-three-field-experiments-software-developers-26pct-tasks-2024.md): Three RCTs at Microsoft, Accenture and a Fortune 100 manufacturer found GitHub Copilot raised completed tasks 26.08% across 4,867 developers, with juniors gaining most.
- [A federal court let a Fair Housing Act claim over State Farm's algorithmic claims handling proceed: Huskey v. State Farm, N.D. Ill., Sept. 11, 2023](https://theinternetninja.com/stories/huskey-v-state-farm-algorithmic-claims-fha-3604b-survives-nd-ill-2023.md): Huskey v. State Farm: on Sept. 11, 2023 a federal court let a Fair Housing Act disparate-impact claim over State Farm algorithmic claims handling proceed.
- [A World Bank RCT in Nigeria: a six-week GPT-4 tutoring program lifted learning by 0.31 standard deviations](https://theinternetninja.com/stories/world-bank-nigeria-generative-ai-tutoring-rct-edo-state-2025.md): World Bank RCT in Nigeria: six-week Microsoft Copilot (GPT-4) tutoring lifted learning 0.31 SD (0.23 SD English), which evaluators put at 1.5 to 2 years of schooling.
- [The Amsterdam Court of Appeal ruled Uber deactivated drivers by automated decision under Article 22 GDPR, rejected its 'human in the loop' as a symbolic act, and ordered disclosure under a 4,000-euro-a-day penalty (2023)](https://theinternetninja.com/stories/amsterdam-court-appeal-uber-ola-drivers-article-22-gdpr-robo-firing-2023.md): On 4 April 2023 Amsterdam's Court of Appeal ruled Uber deactivated drivers by automated decision under Article 22 GDPR, with a 4,000-euro-a-day disclosure penalty.
- [New York's financial regulator investigated the Apple Card algorithm for gender bias and found no unlawful discrimination](https://theinternetninja.com/stories/apple-card-goldman-sachs-nydfs-no-unlawful-discrimination-fair-lending-2021.md): The Apple Card gender bias investigation: New York's DFS analysed ~400,000 applicants and found no unlawful discrimination in Goldman Sachs underwriting.
- [Babylon Health, the AI-first primary-care company, collapsed into bankruptcy in 2023 and sold its UK service for $620,000](https://theinternetninja.com/stories/babylon-health-ai-first-primary-care-spac-collapse-chapter-7-bankruptcy-2023.md): Babylon Health bankruptcy: the AI-first primary-care company collapsed into Chapter 7 in 2023 after a SPAC listing, selling its UK service to eMed for $620,000.
- [Korea's Fair Trade Commission fined Coupang 140 billion won over its search-ranking algorithm: 64,250 own-brand products pinned to the top and 72,614 employee-written reviews in 2024](https://theinternetninja.com/stories/coupang-kftc-search-algorithm-manipulation-pb-products-140bn-won-fine-2024.md): KFTC v. Coupang: Korea's Fair Trade Commission fined Coupang 140 billion won in June 2024 for rigging its search-ranking algorithm to favour its own-brand products.
- [Massachusetts fined Earnest $2.5M over AI underwriting that penalised Black, Hispanic and non-citizen student-loan applicants (2025)](https://theinternetninja.com/stories/earnest-operations-mass-ag-ai-underwriting-fair-lending-2-5m-settlement-2025.md): Massachusetts AG's $2.5M settlement with Earnest Operations over AI underwriting that penalised Black and Hispanic applicants via a Cohort Default Rate variable.
- [The FTC permanently banned FBA Machine over its 'AI-powered' online-store income scheme, with a $15.7M judgment (2025)](https://theinternetninja.com/stories/ftc-fba-machine-passive-scaling-ai-storefront-scheme-15-7m-permanent-ban-2025.md): FTC permanently banned FBA Machine/Passive Scaling over 'AI-powered' storefront income claims, with a $15.7M judgment. Figures quoted verbatim from FTC releases.
- [A Florida federal judge refused to call Character.AI's chatbot output 'speech' and let a wrongful-death product-liability suit proceed](https://theinternetninja.com/stories/garcia-v-character-technologies-ai-chatbot-not-speech-product-liability-2025.md): Garcia v. Character Technologies: a Florida federal judge declined to hold Character.AI's chatbot output is First Amendment speech, letting a wrongful-death suit proceed.
- [Opendoor's algorithmic home-pricing model drove a $928M quarterly loss and a $573M home write-down, then 18% layoffs (Q3 2022)](https://theinternetninja.com/stories/opendoor-ibuying-avm-pricing-algorithm-928m-q3-2022-net-loss-573m-inventory-writedown-18pct-layoffs.md): Opendoor's iBuying automated-valuation pricing model drove a $928M net loss and a $573M home write-down in Q3 2022, followed by 18% layoffs. Verified vs its SEC filing.
- [Pennsylvania made GEICO fix an AI-driven auto-insurance cancellation process the state called unfair and confusing (2026)](https://theinternetninja.com/stories/pennsylvania-ag-geico-ai-underwriting-auto-insurance-cancellation-settlement-2026.md): Pennsylvania AG Dave Sunday made GEICO fix an AI-driven auto-insurance cancellation process in 2026 after an AI tool flagged a policyholder cancelled without notice.
- [South Korea's first AI privacy sanction: the PIPC fined Scatter Lab KRW 103.3 million for training the Iruda chatbot on 9.4 billion KakaoTalk messages](https://theinternetninja.com/stories/scatter-lab-iruda-chatbot-korea-pipc-103m-won-first-ai-privacy-fine-2021.md): Korea's PIPC fined Scatter Lab KRW 103.3 million over the Iruda chatbot in 2021, its first AI sanction under the Personal Information Protection Act.
- [ChatGPT cut Stack Overflow's question volume by about 25% in six months: the peer-reviewed measure of an AI-disrupted knowledge platform, and the 28% layoff that followed](https://theinternetninja.com/stories/stack-overflow-chatgpt-25pct-activity-decline-pnas-nexus-28pct-layoffs-2023.md): Peer-reviewed PNAS Nexus study: ChatGPT cut Stack Overflow posting activity ~25% in six months; Stack Overflow then laid off 28% of staff in October 2023.
- [A Miami federal jury found Tesla's Autopilot defective: $200 million in punitive damages and a ~$243 million judgment in the first U.S. Autopilot death verdict](https://theinternetninja.com/stories/tesla-autopilot-benavides-miami-federal-jury-243m-defective-verdict-2025.md): Benavides v. Tesla: a Miami federal jury found Autopilot defective and 33% liable; the Final Judgment entered $200M in punitive damages, ~$242.5M total.
- [The first pedestrian killed by a self-driving car: NTSB found Uber's automated system detected Elaine Herzberg 5.6 seconds out, never classified her as a pedestrian, and had emergency braking disabled (Tempe, 2018)](https://theinternetninja.com/stories/uber-atg-self-driving-fatality-herzberg-tempe-ntsb-probable-cause-2018.md): NTSB on the first pedestrian death by a self-driving car: Uber's system detected Elaine Herzberg 5.6 seconds before impact but never classified her as a pedestrian.
- [Nature published AlphaFold3 without its code, and it took an open letter and six months to get it released](https://theinternetninja.com/stories/deepmind-alphafold3-nature-no-code-open-letter-github-noncommercial-release-2024.md): AlphaFold3 shipped in Nature without its code. After an open letter and six months, DeepMind released the AlphaFold3 code on GitHub under a noncommercial license.
- [France's CNIL fined Clearview AI the maximum €20 million for facial-recognition scraping, then added a €5.2 million penalty when it did not comply](https://theinternetninja.com/stories/french-cnil-clearview-ai-20m-fine-5-2m-penalty-facial-recognition-gdpr-2022.md): France's CNIL fined Clearview AI the 20 million euro GDPR maximum in 2022 for unlawful facial-recognition scraping, and 5.2 million euro more in 2023 for non-compliance.
- [A court declared a government fraud algorithm unlawful: the Dutch SyRI judgment](https://theinternetninja.com/stories/hague-district-court-syri-welfare-fraud-algorithm-unlawful-article-8-echr-2020.md): On 5 February 2020 the District Court of The Hague ruled the Dutch SyRI welfare-fraud algorithm unlawful for breaching Article 8 ECHR, the right to private life.
- [A benefits algorithm brought down a government: the Dutch childcare-benefits scandal and its EUR 2.75m discrimination fine](https://theinternetninja.com/stories/netherlands-toeslagenaffaire-childcare-benefits-risk-algorithm-dpa-2-75m-fine-2021.md): The Dutch DPA fined the Tax Administration EUR 2.75m for using nationality in a childcare-benefits risk algorithm: the toeslagenaffaire that brought down a cabinet.
- [Texas wins a $1.4B settlement from Meta over facial-recognition biometric capture — 2024](https://theinternetninja.com/stories/texas-ag-meta-1-4b-biometric-facial-recognition-cubi-settlement-2024.md): Texas's $1.4 billion facial-recognition settlement with Meta, 2024: the largest ever from a single-state action, under the CUBI biometric law.
- [Italy's competition authority fined TikTok 10 million euros over its recommendation algorithm: the 'French scar' challenge, algorithmic profiling of minors, and three ByteDance companies fined in solido in 2024](https://theinternetninja.com/stories/tiktok-agcm-french-scar-recommendation-algorithm-10m-fine-minors-2024.md): AGCM v. TikTok: Italy's competition authority fined TikTok 10 million euros in March 2024, finding its algorithm re-proposed the 'French scar' challenge to minors.
- [When a court ruled the computer was wrong: the Post Office Horizon scandal, 39 convictions quashed and an Act of Parliament to overturn the rest](https://theinternetninja.com/stories/uk-post-office-horizon-it-scandal-court-of-appeal-quashes-39-convictions-2021.md): The Post Office Horizon IT scandal: the Court of Appeal quashed 39 of 42 convictions in [2021] EWCA Crim 577 after Horizon was found prone to bugs, errors and defects.
- [Barrows v. Humana: the nH Predict AI coverage-denial suit where breach of contract, good faith and fraud survived a motion to dismiss](https://theinternetninja.com/stories/barrows-v-humana-nh-predict-ai-post-acute-denials-wd-ky-breach-of-contract-survives-2025.md): Barrows v. Humana: on Aug 14, 2025 a federal court let breach-of-contract, good-faith and fraud claims over the nH Predict AI coverage-denial tool survive dismissal.
- [GraphCast: Google DeepMind's AI weather model beat the ECMWF gold standard on 90% of 1,380 targets, and made a 10-day forecast in under a minute](https://theinternetninja.com/stories/google-deepmind-graphcast-outperforms-ecmwf-hres-90pct-1380-targets-2023.md): Google DeepMind's GraphCast AI weather model beat ECMWF's HRES on 90% of 1,380 verification targets and produced a 10-day forecast in under a minute (Science, 2023).
- [Italy's Garante fines Deliveroo €2.5M for its rider-management algorithm — 2021](https://theinternetninja.com/stories/italy-garante-deliveroo-2-5m-rider-management-algorithm-2021.md): Italy's Garante fined Deliveroo €2.5M in 2021 for non-transparent rider-management algorithms and excessive data collection on about 8,000 riders.
- [The Justice Department made Meta rebuild its ad algorithm: a $115,054 maximum penalty, the first FHA algorithmic-bias case, and a Variance Reduction System under court oversight to 2026](https://theinternetninja.com/stories/meta-doj-fair-housing-act-ad-algorithm-vrs-115054-max-penalty-2022.md): DOJ v. Meta: the first Fair Housing Act algorithmic-bias case forced Meta to pay the $115,054 maximum penalty and drop its Special Ad Audience tool.
- [NHTSA's Tesla Autopilot recall: 2,031,220 vehicles fixed over-the-air after regulators found Autosteer's controls insufficient to prevent driver misuse](https://theinternetninja.com/stories/tesla-autopilot-nhtsa-recall-2031220-vehicles-autosteer-controls-misuse-2023.md): Tesla Autopilot recall 23V-838 covered 2,031,220 vehicles after NHTSA found Autosteer's controls insufficient to prevent driver misuse; fixed by an over-the-air update.
- [Tutor CoPilot: a Stanford randomized trial found AI real-time coaching made tutors more effective, with the biggest gains for the weakest tutors](https://theinternetninja.com/stories/tutor-copilot-stanford-fev-rct-4pp-mastery-9pp-lower-rated-tutors-2024.md): Stanford's Tutor CoPilot RCT: AI real-time coaching lifted topic mastery 4 pp, and 9 pp for students of lower-rated tutors, at $20 per tutor a year.
- [Michigan's MiDAS fraud algorithm wrongly accused tens of thousands: 40,195 cases decided by machine, a 93% error rate, and a $20 million settlement](https://theinternetninja.com/stories/michigan-midas-unemployment-fraud-algorithm-40195-cases-93pct-error-20m-settlement-2024.md): Michigan MiDAS: an automated unemployment-fraud system decided 40,195 cases; a state review found 93% wrong, and a $20 million settlement was approved in 2024.
- [The 'jagged frontier' experiment: GPT-4 let 758 BCG consultants finish 12.2% more tasks 25.1% faster, yet made them 19% less likely to be right on a task outside AI's reach](https://theinternetninja.com/stories/bcg-harvard-jagged-frontier-gpt4-consultants-12-2pct-more-tasks-2023.md): A pre-registered study found GPT-4 let 758 BCG consultants do 12.2% more tasks 25.1% faster, yet 19% less likely to be correct outside AI's 'jagged frontier'.
- [JPMorgan Chase scaled its in-house AI assistant, LLM Suite, from zero to 200,000 employees in eight months](https://theinternetninja.com/stories/jpmorgan-chase-llm-suite-genai-200000-employees-eight-months-2025.md): JPMorgan Chase's in-house LLM Suite reached ~200,000 employees in about eight months, with 450+ generative-AI use cases and self-reported productivity gains.
- [AlphaFold: independently ranked the top method in the CASP14 blind assessment (summed z-score 244 vs 91 for second place), scaled to 214M+ predicted structures, and won the 2024 Nobel Prize in Chemistry](https://theinternetninja.com/stories/deepmind-alphafold-casp14-92-4-gdt-independent-top-rank-2024-nobel-chemistry.md): AlphaFold case study: independently ranked #1 in the CASP14 blind assessment (median 92.4 GDT) and won the 2024 Nobel Prize in Chemistry.
- [Bank of Korea (Issue Note 2026-12): AI adoption cut work time 3.8% but the productivity gain is near zero](https://theinternetninja.com/stories/bank-of-korea-2026-ai-adoption-cuts-work-time-3-8pct-productivity-gain-near-zero.md): Bank of Korea Issue Note 2026-12: AI adoption cut work time 3.8 percent, about 1.5 hours a week, yet the productivity gain was near zero.
- [Court of Rome annuls the Garante's €15M ChatGPT fine against OpenAI on one-stop-shop jurisdiction (R.G. 4785/2025, 2026)](https://theinternetninja.com/stories/court-of-rome-annuls-garante-openai-15m-chatgpt-gdpr-fine-jurisdiction-2026.md): Italy's Garante fined OpenAI €15M over ChatGPT in 2024; the Court of Rome annulled the fine in March 2026, ruling the Irish DPC was the GDPR one-stop-shop lead authority.
- [The AI shopping app that ran on people: DOJ and SEC charge Nate's founder over an 'effectively zero percent' automation rate](https://theinternetninja.com/stories/doj-sec-nate-saniger-ai-shopping-app-effectively-zero-percent-automation-40m-securities-wire-fraud-2025.md): DOJ and SEC charged Nate founder Albert Saniger with fraud: the AI shopping app's real automation rate was 'effectively zero percent', run by Philippines contractors.
- [The 'AI-powered' store that lost its investors everything: the FTC banned Automators AI and entered a $21.7M judgment](https://theinternetninja.com/stories/ftc-automators-ai-ecommerce-scheme-cresto-21-7m-judgment-lifetime-ban-2024.md): Automators AI FTC case: operators banned and a $21,765,902.65 judgment entered over an AI e-commerce automation scheme that took $22 million from buyers.
- [GEMA v Suno (LG München I, 42 O 763/25): Munich court rules AI music training and model memorisation infringe copyright](https://theinternetninja.com/stories/gema-suno-lg-muenchen-ai-music-training-copyright-42-o-763-25-2026.md): Landgericht München I (42 O 763/25) ruled Suno's AI music training and memorisation of six protected works infringes copyright; judgment not final.
- [A German court held Google directly liable for its AI Overviews: the Landgericht München I ruled the AI summary is Google's own speech, not a search result (26 O 869/26)](https://theinternetninja.com/stories/landgericht-muenchen-google-ai-overviews-defamation-26-o-869-26-2026.md): Google AI Overviews defamation: a Munich court (LG München I, 26 O 869/26) held Google directly liable for its AI summary as its own attributable content.
- [Apple's $250M Siri settlement: the price of an over-promised Apple Intelligence, in the public record](https://theinternetninja.com/stories/landsheft-v-apple-250m-settlement-apple-intelligence-siri-overpromised-ai-features-2026.md): Apple's $250M Landsheft settlement over an over-promised Apple Intelligence and Siri: preliminary approval pays roughly 36M U.S. devices $25 each, up to $95.
- [Nature Medicine (2026): general-purpose LLMs outperform two deployed specialized clinical AI tools on medical benchmarks](https://theinternetninja.com/stories/nature-medicine-2026-general-purpose-llms-outperform-specialized-clinical-ai-tools-medqa-healthbench-rcq.md): Nature Medicine (2026) benchmark: general-purpose LLMs outperform specialized clinical AI tools OpenEvidence and UpToDate on MedQA, HealthBench and clinical queries.
- [SAVE-O2 AI randomized trial: autonomous oxygen titration kept acutely ill adults in the target range 85% of the time vs 63% under usual care (JAMA Internal Medicine, 2026)](https://theinternetninja.com/stories/save-o2-ai-autonomous-oxygen-titration-jama-85pct-vs-63pct-normoxemia-2026.md): A JAMA Internal Medicine RCT found autonomous closed-loop oxygen titration kept acutely ill adults in the target SpO2 range 85% of the time vs 63% under usual care.
- [The FCC fined Lingo Telecom $1 million for carrying the AI Biden deepfake robocalls that told New Hampshire voters to stay home](https://theinternetninja.com/stories/fcc-lingo-telecom-1m-ai-biden-deepfake-robocall-stir-shaken-election-interference-2024.md): The FCC fined Lingo Telecom $1 million for carrying the AI Biden deepfake robocall that told New Hampshire voters not to vote.
- [An OpenAI reasoning model reanalyzed 376 unsolved rare-disease cases at Boston Children's. Experts confirmed 18 new diagnoses.](https://theinternetninja.com/stories/boston-childrens-manton-openai-o3-deep-research-rare-disease-reanalysis-18-diagnoses-376-cases-nejm-ai-2026.md): AI rare disease diagnosis: OpenAI's o3 reanalyzed 376 unsolved cases at Boston Children's, and experts confirmed 18 new diagnoses, a 4.8% yield.
- [The FTC fines Cox Media Group and two partners $930,000 over an 'Active Listening' AI ad service that never listened](https://theinternetninja.com/stories/ftc-cox-media-group-active-listening-ai-eavesdropping-930000-deceptive-marketing-settlement-2026.md): The FTC fined Cox Media Group, MindSift and 1010 Digital Works $930,000 over the 'Active Listening' AI ad service that used no voice data.
- [Bartz v. Anthropic: a $1.5 billion settlement, the largest in U.S. copyright history, over pirated books used to train Claude](https://theinternetninja.com/stories/bartz-v-anthropic-nd-cal-1-5-billion-ai-copyright-settlement-500000-pirated-books-2026.md): Bartz v. Anthropic: a $1.5 billion settlement, the largest in U.S. copyright history, over pirated books used to train Claude.
- [When an AI denial tool reaches federal court: in Lokken v. UnitedHealth, the breach-of-contract claims over 'nH Predict' survive Medicare preemption](https://theinternetninja.com/stories/lokken-v-unitedhealth-nh-predict-ai-coverage-denials-d-minn-breach-of-contract-survives-preemption-2025.md): Lokken v. UnitedHealth: on Feb 13, 2025 a federal court let breach-of-contract claims over the nH Predict AI coverage-denial tool survive Medicare preemption.
- [A German court says you own what your chatbot says: OLG Hamm holds a clinic liable when its AI invented specialist titles](https://theinternetninja.com/stories/olg-hamm-aesthetify-ai-chatbot-fabricated-facharzt-titles-uwg-liability-cease-desist-2026.md): OLG Hamm chatbot liability: a German court held a company liable for its AI chatbot false specialist titles under UWG, with a 250,000 euro per-violation fine ceiling.
- [Dutch DPA fines Uber €824.99M for automated driver deactivation without human review](https://theinternetninja.com/stories/uber-autoriteit-persoonsgegevens-824-99m-gdpr-automated-driver-deactivation-2026.md): On 21 Aug 2026 the Dutch DPA fined Uber €824.99M for automated driver deactivation without human review, a prohibited fully automated GDPR decision.
- [The UK government banks £45bn a year on AI and digital efficiencies its own auditor says are not yet evidenced](https://theinternetninja.com/stories/uk-nao-government-workforce-planning-45bn-ai-digital-savings-claim-2026.md): The UK's National Audit Office logs a £45bn-a-year AI and digital savings expectation, then flags that departments never showed how they derived it.
- [A VA watchdog found errors in nearly all 8,100 automated survivors' death-benefit decisions it reviewed, with $2.7M in improper payments](https://theinternetninja.com/stories/va-oig-automated-survivors-death-benefit-decisions-8000-errors-2-7m-improper-payments-2026.md): VA OIG audit of automated survivors' death-benefit (DIC) decisions found deficiencies in at least 8,000 of ~8,100 reviewed and $2.7M in improper payments.
- [SEC 'AI-washing' enforcement: Rimar Capital and owner Itai Liptz pay $310,000 for an AI trading platform that didn't exist](https://theinternetninja.com/stories/sec-ai-washing-rimar-capital-itai-liptz-310000-civil-penalties-2024.md): SEC AI-washing enforcement: Rimar Capital and owner Itai Liptz paid $310,000 over an AI trading platform that never existed (SEC Order 33-11316, 2024).
- [A Connecticut federal judge fines a lawyer $3,500 and refers him for discipline over AI-fabricated legal authorities](https://theinternetninja.com/stories/barteca-v-tacobarn-d-conn-oliver-miller-3500-ai-fabricated-citations-grievance-referral-2026.md): Barteca v tacobarn: a Connecticut federal judge fined lawyer Hilary Miller $3,500 and referred him for discipline over AI-fabricated legal authorities.
- [Pooja Ramesh Singh v. J&K Bank [2026] INSC 668: India's Supreme Court sets aside a tribunal's own AI-hallucinated 'precedents', 'no decision in the eyes of the law'](https://theinternetninja.com/stories/pooja-ramesh-singh-v-jk-bank-supreme-court-india-insc-668-nclt-ai-hallucinated-precedents-set-aside-2026.md): Pooja Ramesh Singh v. J&K Bank (2026 INSC 668): India's Supreme Court set aside NCLT/NCLAT orders built on AI-hallucinated 'precedents'. No fine; zero-tolerance rule.
- [Re A, B, C, D [2026] EWFC 71 (B): a family court names an unregistered barrister who put four AI-hallucinated authorities before it](https://theinternetninja.com/stories/re-abcd-ewfc-71-b-bournemouth-parsons-unregistered-barrister-ai-four-fabricated-cases-named-2026.md): Re A, B, C, D [2026] EWFC 71 (B): Bournemouth Family Court named unregistered barrister Layla Parsons for four AI-fabricated citations. No fine.
- [Connecticut's high court, for the first time, sanctions a lawyer for ChatGPT-hallucinated citations](https://theinternetninja.com/stories/ct-supreme-court-tov-realty-kosel-gottlieb-glg-law-chatgpt-hallucinated-citations-1000-donation-cle-first-impression-2026.md): Connecticut's Supreme Court sanctioned Attorney Ian Gottlieb and GLG Law for ChatGPT-hallucinated citations: six hours of CLE and $1,000 each to the CT Bar Institute.
- [An AI-drafted petition, a 'wrong draft' excuse, and a referral to the Florida Bar: JMOR Properties v. Artist Alley Townhomes](https://theinternetninja.com/stories/jmor-properties-v-artist-alley-fla-4th-dca-leff-florida-bar-referral-ai-drafted-certiorari-fabricated-citations-2026.md): JMOR Properties v. Artist Alley: Florida's Fourth DCA referred attorney Barry Leff to the Florida Bar over an AI-drafted certiorari petition of false citations.
- [The Third Circuit's first AI-hallucination sanction: a public reprimand for unverified, AI-generated citations](https://theinternetninja.com/stories/mccarthy-v-dea-3d-cir-24-2704-ai-fabricated-citations-public-reprimand-first-impression-2026.md): Third Circuit's first AI-hallucination sanction: Daniel Pallen reprimanded after 7 of 8 AI-generated DEA case summaries were inaccurate and one case did not exist.
- [Spain's AEPD fines Mercadona €2,520,000 over its store-entrance facial-recognition system](https://theinternetninja.com/stories/spain-aepd-fines-mercadona-2520000-facial-recognition-biometric-ai-gdpr-sanction-2021.md): Spain's AEPD fined Mercadona €2,520,000 over its store-entrance facial-recognition system, prohibiting all such biometric processing under GDPR Article 9.
- [Rechtbank Amsterdam raises a costs award by €2,315.50 after a fabricated DSM/Fox citation the losing party blamed on AI hallucination](https://theinternetninja.com/stories/rechtbank-amsterdam-rbams-2026-8258-dsm-fox-fabricated-citation-2315-euro-proceskosten-ai-hallucination-2026.md): Rechtbank Amsterdam (ECLI:NL:RBAMS:2026:8258) added €2,315.50 to a costs award over a fabricated DSM/Fox citation the losing party blamed on AI hallucination.
- [Quebec's appeal court ordered a self-represented litigant to pay $1,000 in costs for filing four ChatGPT-fabricated court decisions](https://theinternetninja.com/stories/a-p-v-s-k-2026-qcca-1092-quebec-court-of-appeal-chatgpt-fabricated-case-law-1000-cad-costs-self-represented-2026.md): Quebec Court of Appeal (2026 QCCA 1092) ordered a self-represented appellant to pay $1,000 costs for filing four ChatGPT-fabricated court decisions.
- [Brazil's ANPD halts Meta's use of personal data to train generative AI (2024)](https://theinternetninja.com/stories/brazil-anpd-halts-meta-generative-ai-training-preventive-measure-50k-daily-fine-2024.md): Brazil's ANPD halted Meta training generative AI on personal data (2 July 2024) under a R$50,000/day fine; Meta then suspended its AI tools in Brazil.
- [Dutch DPA fines the Tax Administration €2.75M for unlawful algorithmic nationality profiling](https://theinternetninja.com/stories/dutch-dpa-fines-tax-authority-2-75m-for-unlawful-algorithmic-risk-profiling-in-c.md): The Dutch DPA fined the Tax Administration €2.75M in 2021 for a self-learning algorithm that used applicants nationality to auto-flag them for risk review.
- [Hungary's NAIH halts a bank's AI emotion analysis of call recordings, HUF 250,000,000 fine (2022)](https://theinternetninja.com/stories/hungarian-naih-budapest-bank-250m-huf-fine-unlawful-ai-emotion-voice-analysis-2022.md): Hungary's NAIH fined Budapest Bank HUF 250,000,000 (about EUR 670,000) and ordered it to stop AI emotion analysis of call recordings: a record GDPR fine (2022).
- [Italy's Garante blocks DeepSeek: limitation on processing of Italian users' data (2025)](https://theinternetninja.com/stories/italy-garante-blocks-deepseek-ai-limitation-processing-italian-users-data-2025.md): Italy's Garante blocked DeepSeek on 30 Jan 2025, ordering an urgent limitation on processing of Italian users' data and opening an investigation.
- [Italy's Garante fines Foodinho (Glovo) for its rider-management algorithm: €2.6M in 2021, €5M in 2024](https://theinternetninja.com/stories/italy-garante-fines-foodinho-2-6m-for-rider-management-algorithm-2021.md): Italy's Garante fined Foodinho (Glovo) €2.6M in 2021 and €5M in 2024 over its rider-management algorithm, ordering changes and banning facial recognition.
- [Spain's AEPD orders Worldcoin (Tools for Humanity) to halt biometric iris-scanning data processing (2024)](https://theinternetninja.com/stories/spain-aepd-orders-worldcoin-tools-for-humanity-to-halt-biometric-iris-scanning-a.md): Spain's AEPD ordered Worldcoin (Tools for Humanity) to halt iris-scanning data processing on 6 Mar 2024; the National Court refused to suspend it.
- [Adobe updated its Terms of Use after an AI-training backlash to state it won't train generative AI on customer content](https://theinternetninja.com/stories/adobe-terms-of-use-ai-training-clarification-added-after-backlash-2024.md): Adobe updated its Terms of Use in June 2024 after an AI-training backlash, adding a statement that it does not train generative AI on customer content.
- [After a December 2023 backlash over a default-on 'third-party AI' setting, Dropbox said customer data is never used to train OpenAI's models and is deleted within 30 days](https://theinternetninja.com/stories/dropbox-third-party-ai-openai-toggle-backlash-data-not-used-to-train-models-2023.md): After a December 2023 backlash over a default-on 'third-party AI' toggle, Dropbox said customer data is never used to train OpenAI's models and is deleted in 30 days.
- [HubSpot reversed a July 2026 terms-of-service change that would have pooled customer data and allowed AI-model training, after customer backlash](https://theinternetninja.com/stories/hubspot-contact-discovery-data-enrichment-ai-terms-reversed-after-backlash-2026.md): HubSpot's July 2026 terms of service reversal: after backlash over Contact Discovery and AI model training, CPTO Duncan Lennox said 'We made a mistake.'
- [Humane discontinued the Ai Pin and bricked shipped devices; HP bought its AI assets for $116M](https://theinternetninja.com/stories/humane-ai-pin-discontinued-hp-116m-acquisition-devices-bricked-2025.md): HP acquired Humane's Cosmos AI platform and 300+ patents for $116M; the Ai Pin was discontinued and shipped devices bricked on Feb 28, 2025.
- [A US federal court let a nationwide age-discrimination case against Workday's AI hiring tool proceed as a collective action (May 2025)](https://theinternetninja.com/stories/mobley-v-workday-nd-cal-ai-hiring-tools-age-discrimination-nationwide-adea-collective-certified-2025.md): Mobley v. Workday: a US federal court granted preliminary ADEA collective certification on May 16, 2025, letting a nationwide AI hiring discrimination lawsuit proceed.
- [Mozilla rewrote Firefox's Terms of Use after backlash, saying the change wasn't about using data for AI](https://theinternetninja.com/stories/mozilla-firefox-terms-of-use-rewritten-after-backlash-not-using-data-for-ai-2025.md): Mozilla rewrote Firefox's Terms of Use after backlash in early 2025, saying the new data license wasn't for training AI. The before/after clauses, verified.
- [After a user backlash, Slack updated its AI 'privacy principles' over training ML models on customer data (May 2024)](https://theinternetninja.com/stories/slack-ai-privacy-principles-updated-after-user-backlash-ml-training-opt-out-2024.md): Slack AI privacy principles were updated in May 2024 after a user backlash over training ML models on customer data, opt-out by default and escapable only by email.
- [X suspended EU/EEA user-data training for Grok after Ireland's DPC took urgent High Court action](https://theinternetninja.com/stories/x-suspends-eu-user-data-training-for-grok-after-irish-dpc-high-court-action-2024.md): Ireland's DPC used an urgent Section 134 High Court application to make X suspend EU/EEA user-data training for Grok on 8 August 2024.
- [After backlash, Zoom reversed its terms of service so it will not use customer content to train AI (August 2023)](https://theinternetninja.com/stories/zoom-terms-of-service-ai-training-customer-content-reversed-after-backlash-2023.md): After a 2023 backlash over its terms of service, Zoom updated section 10.4 and said it will not use customer content to train AI without consent.
- [WeTransfer removed the machine-learning clause from its Terms of Service after user backlash (July 2025)](https://theinternetninja.com/stories/wetransfer-ai-training-clause-terms-removed-after-backlash-2025.md): WeTransfer removed the machine-learning clause from its Terms of Service in July 2025 after a backlash over AI training on user files, then reversed it.
- [Chicago Sun-Times ran an AI-generated summer reading list where only 5 of 15 books were real](https://theinternetninja.com/stories/chicago-sun-times-ai-fabricated-summer-reading-list-10-of-15-fake-books-king-features-2025.md): Chicago Sun-Times ran an AI-generated summer reading list where only 5 of 15 books were real; King Features licensed it and the CEO acknowledged the error.
- [Google scaled back AI Overviews after it told users to put glue on pizza and eat rocks](https://theinternetninja.com/stories/google-ai-overviews-glue-pizza-eat-rocks-scaled-back-2024.md): Google scaled back AI Overviews after it told users to glue pizza and eat rocks: the errors, Google's response, and the fixes it shipped.
- [Microsoft capped its 'unhinged' new Bing AI chat at 5 turns per session and 50 per day](https://theinternetninja.com/stories/microsoft-bing-ai-chat-capped-5-turns-per-session-50-per-day-unhinged-responses-2023.md): Microsoft capped its 'unhinged' new Bing AI chat at 5 turns per session and 50 per day in February 2023, then loosened the limits to 6 and 60 days later.
- [Microsoft's "Tay" chatbot was pulled offline within ~16 hours after learning to be racist](https://theinternetninja.com/stories/microsoft-tay-ai-chatbot-taken-offline-racist-tweets-16-hours-2016.md): Microsoft Tay chatbot went offline within ~16 hours in 2016 after users taught it racist tweets: 95,000+ tweets, then a Microsoft apology for a critical oversight.
- [Zillow shut down its algorithmic home-buying business after a $304M write-down and 25% layoffs](https://theinternetninja.com/stories/zillow-offers-ibuying-algorithm-shutdown-304m-writedown-25pct-layoffs-2021.md): Zillow Offers shutdown: Zillow wound down its iBuying business after a $304M write-down and a 25% workforce cut when its Zestimate algorithm mispriced homes.
- [ADGM Court (Heath J.): AED 282,508 indemnity costs against MIO Legal Consultants for AI-fabricated authorities in a defence pleading](https://theinternetninja.com/stories/arabyads-v-alam-adgm-cfi-heath-282508-aed-mio-legal-ai-fabricated-authorities-indemnity-costs-2025.md): ADGM Court ordered MIO Legal Consultants to pay AED 282,508 indemnity costs for AI-fabricated authorities in Arabyads v Alam, a reckless breach of the Rules of Conduct.
- [Tribunale di Siracusa n. 338/2026: ~€30,206 in awards after four Cassation citations with fabricated holdings, court infers unverified generative-AI use](https://theinternetninja.com/stories/italy-tribunale-siracusa-338-2026-spitaleri-30206-euro-ai-fabricated-citations-lite-temeraria.md): Tribunale di Siracusa 338/2026: an Italian court imposed 30,206 euros in three awards for AI-fabricated citations under art. 96 c.p.c. lite temeraria.
- [Arizona federal court: attorney sanctioned for AI-fabricated citations, only 5 to 7 of 19 cases were real, pro hac vice revoked, no fine](https://theinternetninja.com/stories/mavy-v-commissioner-social-security-d-ariz-bachus-bam-12-of-19-ai-fabricated-citations-pro-hac-vice-revoked-wsba-referral-2025.md): Mavy v. Commissioner (D. Ariz., 2025): only 5 to 7 of 19 citations were real; the court revoked Maren Bam's pro hac vice status, no monetary fine.
- [Miller v. Regions Bank: an Alabama attorney deleted his ChatGPT account to hide AI-fabricated citations, and lost far more than the case](https://theinternetninja.com/stories/miller-v-regions-bank-nd-ala-mooty-harp-chatgpt-deleted-account-six-month-suspension-public-reprimand-2026.md): Miller v. Regions Bank: an Alabama federal judge suspended attorney H. Gregory Harp six months for AI-fabricated citations and deleting his ChatGPT account.
- [A federal appeals court fines a lawyer $2,500 for an AI-drafted brief, even though she used purpose-built legal-AI tools](https://theinternetninja.com/stories/fletcher-v-experian-5th-cir-hersh-2500-ai-fabricated-citations-frap-46c-inherent-authority-2026.md): Fletcher v. Experian: the Fifth Circuit fined lawyer Heather Hersh $2,500 under FRAP 46(c) for an AI-drafted brief with 16 fabricated quotations.
- [A Massachusetts judge fines a lawyer $2,000 for filing briefs full of AI-invented cases, the state's first](https://theinternetninja.com/stories/smith-v-farwell-mass-super-davis-2000-ai-fabricated-citations-2024.md): Smith v. Farwell: a Massachusetts judge fined a lawyer $2,000 under Rule 11 for briefs built on AI-hallucinated, wholly-fictitious case citations. The state's first.
- [An Illinois appeals court fined attorney Mason Cole $15,000 for ChatGPT-hallucinated citations](https://theinternetninja.com/stories/scott-v-illinois-hrc-ill-app-1st-van-tine-cole-15000-chatgpt-fabricated-citations-rule-375-2026.md): Scott v. Illinois Human Rights Comm'n: an Illinois appeals court fined attorney Mason Cole $15,000 for 10 ChatGPT-hallucinated citations and referred him to the ARDC.
- [Lichfield v. Kubler: 10th Circuit finds AI-fabricated citations but declines to sanction](https://theinternetninja.com/stories/lichfield-v-kubler-10th-cir-federico-published-ai-fabricated-citations-no-sanction-notice-of-errata-2026.md): Lichfield v. Kubler: the Tenth Circuit found AI-fabricated citations in the opening brief, held it non-compliant with Rule 46.5, then declined to sanction.
- [Ontario Law Society Tribunal orders a lawyer to pay $31,150 in costs after Grok-generated fake citations](https://theinternetninja.com/stories/mazaheri-v-law-society-ontario-onlsth-112-31150-costs-grok-ai-fabricated-citations-2026.md): Mazaheri v Law Society of Ontario (2026 ONLSTH 112): a lawyer ordered to pay $31,150 in costs after filing Grok-generated fake citations.
- [OTG New York v. Ottogi America: a New Jersey federal court fines a lawyer $3,000 for a reply brief built on AI-hallucinated citations (2025)](https://theinternetninja.com/stories/otg-v-ottogi-dnj-almonte-cho-3000-ai-fabricated-citations-rule-11-sanction-2025.md): OTG v. Ottogi: a New Jersey federal court sanctioned attorney Sukjin Henry Cho $3,000 under Rule 11 for a reply brief built on AI-hallucinated citations.
- [Torres Campos v. Munoz: a published California appellate opinion sanctions a lawyer $5,000 for citing AI-hallucinated cases in a dog-custody fight (2026)](https://theinternetninja.com/stories/torres-campos-v-munoz-cal-app-4th-buchanan-5000-ai-fabricated-citations-dog-custody-2026.md): Torres Campos v. Munoz: a published California appellate opinion sanctioned attorney Roxanne Chung Bonar $5,000 for citing AI-hallucinated cases in a dog-custody appeal.
- [Air Canada held liable for its website chatbot's bad advice: tribunal orders C$812.02 (Moffatt v. Air Canada, 2024 BCCRT 149)](https://theinternetninja.com/stories/air-canada-chatbot-bereavement-misinfo-tribunal-orders-812-cad-negligent-misrepresentation-2024.md): The Air Canada chatbot lawsuit, verified: in Moffatt v. Air Canada a BC tribunal ordered C$812.02 for the bot's bad bereavement-fare advice.
- [A federal court prices an unchecked AI legal workflow: $3,000 sanctions each for the attorneys behind an AI-drafted brief with nearly thirty defective citations](https://theinternetninja.com/stories/coomer-v-lindell-mypillow-attorneys-ai-brief-30-defective-citations-3000-sanctions-each-2025.md): Coomer v. Lindell: a federal judge sanctioned Mike Lindell's lawyers $3,000 each under Rule 11 for an AI-drafted brief with nearly thirty defective citations.
- [€30.5 million for an illegal AI face database: the Dutch DPA's final, unappealable GDPR fine on Clearview AI](https://theinternetninja.com/stories/dutch-dpa-clearview-ai-gdpr-30-5m-fine-illegal-facial-recognition-database-2024.md): The Dutch DPA fined Clearview AI €30.5 million for an illegal facial-recognition database of 30 billion scraped faces. The final GDPR decision, verified.
- [iTutorGroup pays $365,000 after the EEOC alleges its application software auto-rejected 200+ older applicants, the settlement the agency's Chair tied to its AI enforcement push](https://theinternetninja.com/stories/eeoc-itutorgroup-ai-hiring-age-screening-365k-consent-decree-2023.md): EEOC v. iTutorGroup: a $365,000 consent decree after application software auto-rejected 200+ applicants over 55 or 60, called the EEOC's first AI hiring case.
- [FTC final order: the 'AI-powered' web-accessibility widget that promised any-website WCAG compliance pays $1,000,000](https://theinternetninja.com/stories/ftc-accessibe-accesswidget-ai-web-accessibility-1m-monetary-relief-final-order-2025.md): FTC accessiBe order: accessWidget's AI claimed WCAG compliance in 48 hours; the FTC found the claim unsubstantiated and required a $1,000,000 payment.
- [Federal court enters $25M judgment and permanent ban against Ascend Ecom after FTC alleges its 'AI-powered' passive-income storefronts earned clients virtually nothing](https://theinternetninja.com/stories/ftc-ascend-ecom-ai-business-opportunity-25m-judgment-court-ordered-ban-2025.md): Ascend Ecom FTC case: the court entered a $25M judgment and a permanent business-opportunity ban over an 'AI-powered' passive-income scheme.
- [FTC orders DoNotPay's 'world's first robot lawyer' to pay $193,000 and notify subscribers over unsubstantiated AI claims](https://theinternetninja.com/stories/ftc-donotpay-ai-robot-lawyer-193k-monetary-relief-final-order-2025.md): FTC final order C-4812 makes DoNotPay's 'world's first robot lawyer' pay $193,000, notify 2021 to 2023 subscribers, and drop unsubstantiated AI lawyer claims.
- [FTC v. Evolv: the AI scanner marketed as detecting all weapons missed a knife used in a school stabbing, and a federal court entered a permanent injunction](https://theinternetninja.com/stories/ftc-evolv-ai-weapons-detection-court-stipulated-injunction-k12-cancel-right-2024.md): FTC v. Evolv: the FTC alleged the AI weapons scanner in 800+ schools overpromised; a federal court entered a permanent injunction with a K-12 right to cancel.
- [FTC × IntelliVision: a 'zero bias, millions of faces' AI claim, measured against NIST](https://theinternetninja.com/stories/ftc-intellivision-facial-recognition-zero-bias-claims-20-year-consent-order-2025.md): IntelliVision claimed zero-bias facial recognition on millions of faces. FTC order C-4809 (2025): about 100,000 people used, no top-100 NIST rank, 20-year ban.
- [Federal court bans Rite Aid from AI facial-recognition surveillance for five years after FTC alleges thousands of false-positive matches](https://theinternetninja.com/stories/ftc-rite-aid-ai-facial-recognition-five-year-ban-court-ordered-2024.md): FTC v. Rite Aid: a court-entered stipulated order bans Rite Aid from AI facial recognition for five years after thousands of alleged false positives.
- [FTC final order: the '98% accurate' AI Content Detector was an unmodified student model, and its own published data showed 53.2% on real-world text](https://theinternetninja.com/stories/ftc-workado-ai-content-detector-98pct-claimed-53pct-actual-accuracy-final-order-2025.md): FTC order C-4822: Workado's AI Content Detector claimed 98.3% accuracy; the FTC record shows 53.2% on the real-world text it was sold to catch.
- [The €15 million ChatGPT fine that didn't survive: Italy's Garante fined OpenAI, the Court of Rome annulled it](https://theinternetninja.com/stories/italy-garante-openai-chatgpt-15m-gdpr-fine-annulled-court-of-rome-4153-2026.md): Italy's Garante fined OpenAI EUR 15M over ChatGPT data practices; the Court of Rome annulled it in judgment 4153/2026 on one-stop-shop jurisdiction, never collected.
- [A $2.275M class settlement takes the algorithm's score away: Louis v. SafeRent and the five-year rollback of tenant-screening scores for voucher applicants](https://theinternetninja.com/stories/louis-v-saferent-algorithmic-tenant-screening-2-275m-class-settlement-dmass-2024.md): Louis v. SafeRent: a $2.275M class settlement and a five-year ban on tenant-screening scores for housing-voucher applicants in D. Mass.
- [The canonical AI-hallucination sanction: $5,000, jointly and severally, for the lawyers who filed six ChatGPT-fabricated cases in Mata v. Avianca](https://theinternetninja.com/stories/mata-v-avianca-chatgpt-fake-citations-5000-sanction-rule-11-sdny-2023.md): Mata v. Avianca: Judge Castel imposed a $5,000 Rule 11 sanction on Steven Schwartz and Peter LoDuca for six ChatGPT-fabricated case citations in the SDNY.
- [SEC AI-washing order: Presto's drive-thru 'voice AI' needed a human on every order](https://theinternetninja.com/stories/sec-ai-washing-presto-automation-voice-ai-drive-thru-human-agents-cease-and-desist-2025.md): SEC AI-washing order: Presto Automation's drive-thru voice AI needed off-site human agents on nearly every order while filings claimed it eliminated human order taking.
- [SEC AI washing enforcement: Delphia and Global Predictions pay $400,000 for AI claims they couldn't back](https://theinternetninja.com/stories/sec-ai-washing-delphia-global-predictions-400k-civil-penalties-2024.md): SEC AI-washing enforcement: Delphia and Global Predictions paid $400,000 in 2024 for AI claims they could not back, under SEC orders IA-6573 and IA-6574.
- [A federal court fined a public defender $1,500 for a fake citation, then declined to say AI wrote it](https://theinternetninja.com/stories/us-v-hayes-federal-defender-1500-sanction-fictitious-citation-court-declines-to-find-ai-use-2025.md): United States v. Hayes: a federal court fined public defender Andrew Francisco $1,500 for a fake citation but declined to find AI wrote it.
- [John Deere See & Spray: independent university trials cut post-emergence herbicide 43% to 62%](https://theinternetninja.com/stories/john-deere-see-spray-ai-spot-spraying-cuts-herbicide-use-43-59-independently-con.md): Independent university trials of John Deere See & Spray report 43% to 62% post-emergence herbicide savings in soybean, with a documented resistance tradeoff.
- [Leiske v. Kidd (Del. Ch.): one of Delaware's oldest firms — Richards, Layton & Finger — ordered to show cause over GenAI-hallucinated citations. No sanction yet.](https://theinternetninja.com/stories/leiske-v-kidd-del-ch-lori-will-richards-layton-finger-rollo-genai-hallucinated-citations-rule-11-show-cause-2026.md): Leiske v. Kidd: Delaware Chancery ordered Richards, Layton & Finger and Richard Rollo to show cause over GenAI-hallucinated citations. No sanction imposed.
- [Romero v. Corona Investments: the Seventh Circuit found 'an astonishing number of hallucinated citations' — and declined to sanction anyway](https://theinternetninja.com/stories/romero-v-corona-investments-7th-cir-bach-ai-hallucinated-citations-admonished-no-sanction-511a-2026.md): Romero v. Corona Investments (7th Cir., 2026): the court found 'hallucinated citations' in Paul Bach's brief but declined to sanction; the dissent would have.
- [Akerlund v. Atlas Air: a federal appeals court refers a former Florida lawmaker to its lawyer-conduct committee over AI-hallucinated citations, filed twice](https://theinternetninja.com/stories/akerlund-v-atlas-air-11th-cir-sabatini-ai-fabricated-hallucinated-citations-committee-referral-2026.md): The Eleventh Circuit faulted Anthony Sabatini in Akerlund v. Atlas Air for AI-hallucinated citations and referred him for discipline. No fine.
- [The UK's DWP fraud model: a machine-learning system that saved an estimated 4.4 million pounds on Universal Credit advances](https://theinternetninja.com/stories/dwp-ml-fraud-model-universal-credit.md): DWP machine learning fraud case study: a model flagging fraudulent Universal Credit advances saved an estimated 4.4 million pounds, per the NAO.
- [Meta pulls Instagram's Muse Image @-mention feature three days after launch](https://theinternetninja.com/stories/meta-instagram-muse-image-ai-feature-removed-three-days-nonconsensual-backlash-2026.md): Meta Muse Image, the Instagram AI feature that let anyone generate images from public accounts, was pulled three days after its July 2026 launch.
- [Coinbase: AI spend cut nearly in half, and the biggest lever was caching, not cheaper models](https://theinternetninja.com/stories/coinbase-ai-spend-open-weight-routing.md): Coinbase AI spend case study: the bill fell nearly 50% while usage rose. The biggest lever was LibreChat caching, 5% to 60%, not the open-weight models.
- [Notion and Decagon: an AI support agent, 34% faster resolution and 2x deflection](https://theinternetninja.com/stories/decagon-notion-ai-cx.md): Notion Decagon AI case study: 34% faster ticket resolution, 2x deflection, 3.4% ask-for-human rate. Vendor figures, CX head named and quoted.
- [Delivery Hero and n8n: account-lockout recovery automated in a single workflow](https://theinternetninja.com/stories/delivery-hero-n8n-it-ops-automation.md): Delivery Hero n8n case study: account-lockout recovery cut from 35 to 20 minutes, 200 hours a month returned, deployed in 5 hours. Director on record.
- [JPMorgan's COIN: the famous 360,000-hours figure, and why it was never independently measured](https://theinternetninja.com/stories/jpmorgan-coin-contract-intelligence.md): JPMorgan COIN case study: the contract intelligence AI, the famous 360,000 hours figure, and why that number was never independently measured.
- [Klarna's AI customer-service assistant: the 2024 numbers and the 2025 walk-back](https://theinternetninja.com/stories/klarna-ai-assistant-customer-service.md): Klarna AI assistant case study, verified: two-thirds of chats in month one, the 700-agent equivalence explained, and the 2025 walk back that rehired humans.
- [Mango's first fully AI-generated campaign: Sunset Dream, 95 markets, and no public performance number](https://theinternetninja.com/stories/mango-ai-generated-fashion-campaign.md): Mango AI generated campaign, verified: the Sunset Dream Teen line, a model trained on real garment photos, live in 95 markets, no public results figure.
- [Moderna's company-wide ChatGPT Enterprise rollout: 750 custom GPTs, from a vendor case study](https://theinternetninja.com/stories/moderna-openai-chatgpt-enterprise.md): Moderna ChatGPT Enterprise case study: 750 custom GPTs in two months, 120 weekly conversations per user, and where the OpenAI-published figures fall short.
- [Octopus Energy's 'Magic Ink': AI that drafts support emails, with no job cuts](https://theinternetninja.com/stories/octopus-energy-kraken-ai-support.md): Octopus Energy AI customer service case study: Kraken's Magic Ink drafted replies, handled 34% of queries by April 2023, a human reviewing every send.
- [Omega Healthcare and UiPath: document automation across billing, coding, and payer correspondence](https://theinternetninja.com/stories/omega-healthcare-uipath-document-automation.md): Omega Healthcare UiPath case study: 40% less documentation time, 50% faster turnaround, 99.5% accuracy, and two hours-saved figures shown by their dates.
- [Otto's autonomous stock ordering: a deep-learning system that buys inventory on its own](https://theinternetninja.com/stories/otto-germany-ai-stock-replenishment.md): Otto AI stock replenishment case study: Blue Yonder deep learning at 90% forecast accuracy, 200,000 auto-ordered items a month, surplus stock down a fifth.
- [TSXG Galicia: Spain's High Court of Justice of Galicia fines a lawyer €1,800 for an appeal built on 24 AI-fabricated citations, and sizes the fine to a legal-AI subscription](https://theinternetninja.com/stories/tsxg-galicia-recurso-5472-abogado-1800-euros-ia-24-citas-fabricadas-buena-fe-procesal-2026.md): TSXG Galicia fined a lawyer €1,800 under article 75.4 LJS for an appeal with 24 AI-fabricated citations, and sized the fine to a legal-AI subscription.
- [Zalando's 2018 'algorithms replace 250 marketing jobs': the headline, and the hiring plan it left out](https://theinternetninja.com/stories/zalando-ai-marketing-roles-2018.md): Zalando AI marketing jobs, the myth checked: the 2018 plan to replace 200 to 250 roles with algorithms sat inside a plan to hire 2,000 more people.
- [Landberg v. City of New York: an AI-fabricated appellate brief drew $10,500 in sanctions](https://theinternetninja.com/stories/landberg-v-city-of-new-york-ny-appellate-second-dept-ai-fabricated-citations-10500-sanction-2026.md): Landberg v City of New York: a NY appellate court sanctioned attorney Michael Sanders $8,000 and his firm $2,500 for an AI-fabricated brief.
- [Italy's Supreme Court fines an appellant €5,000 for AI-hallucinated case citations](https://theinternetninja.com/stories/italy-cassazione-23006-2026-5000-euro-ai-fabricated-citations-diligenza-digitale.md): Cassazione 23006/2026: Italy's Supreme Court fined an appellant 5,000 euros for AI-hallucinated citations, ruling the signing lawyer stays responsible.
- [Barber v. Morawa: Michigan's first published appellate opinion sanctioning a lawyer for AI-fabricated citations](https://theinternetninja.com/stories/barber-v-morawa-mich-ct-app-first-published-ai-fabricated-citations-vexatious-appeal-sanction-grievance-referral-2026.md): Barber v. Morawa: Michigan's first published appellate opinion sanctioning an attorney for AI-fabricated citations, with a Grievance Commission referral.
- [Withers v. City of Aberdeen: a federal court sanctions attorneys on both sides $8,000 for AI-hallucinated citations](https://theinternetninja.com/stories/withers-v-city-of-aberdeen-nd-miss-both-sides-ai-hallucinated-citations-8000-rule-11-sanctions-2026.md): Withers v. City of Aberdeen: a federal judge sanctioned attorneys on both sides $8,000 for AI-hallucinated citations and disqualified local counsel.
- [Alabama Supreme Court dismisses appeal, fines lawyer $17,200 for AI-'hallucinated' citations](https://theinternetninja.com/stories/ibach-v-stewart-alabama-supreme-court-hall-17200-ai-fabricated-23-citations-frivolous-appeal-dismissed-2026.md): Ibach v. Stewart: the Alabama Supreme Court dismissed the appeal as frivolous and fined attorney W. Perry Hall $17,200 for AI-hallucinated citations.
- [Rodney v Gee'z Micro Bar: a County Court refers two solicitors to the SRA over AI-suspected fake citations](https://theinternetninja.com/stories/rodney-v-geez-micro-bar-county-court-dudley-grimshaw-aml-legal-sra-referral-ai-fabricated-citations-2026.md): Rodney v Gee'z Micro Bar: HHJ Grimshaw referred AML Legal solicitors to the SRA over AI-suspected fabricated citations at the County Court at Dudley. No fine imposed.
- [United States v. Farris: a federal appeals court denies a lawyer his pay and removes him from the case over AI-fabricated quotations from Westlaw's CoCounsel](https://theinternetninja.com/stories/us-v-farris-6th-cir-howe-cocounsel-ai-fabricated-quotations-cja-compensation-denied-removed-2026.md): United States v. Farris: the Sixth Circuit denied attorney Steven Howe his Criminal Justice Act pay and removed him over AI-fabricated CoCounsel quotations. No fine.
- [Italy's Supreme Court fines an appellant EUR 3,000 for AI-hallucinated citations — Cass. pen. Sez. 7 Ord. 11431/2026](https://theinternetninja.com/stories/italy-cassazione-11431-2026-sez7-penale-3000-euro-ai-fabricated-citations-diligenza-digitale.md): Cassazione 11431/2026: Italy's Supreme Court declared an appeal inadmissible and ordered a 3,000 euro payment over AI-hallucinated case citations.
- [Prososki v. Regan: Nebraska Supreme Court strikes an appellate brief and refers counsel for discipline over fictitious, allegedly AI-generated citations](https://theinternetninja.com/stories/prososki-v-regan-nebraska-supreme-court-lake-ai-fictitious-citations-brief-struck-appeal-dismissed-discipline-referral-2026.md): Prososki v. Regan, verified: the Nebraska Supreme Court struck an appellate brief and referred counsel for discipline over fictitious, allegedly AI-generated citations.
- [Whiting v. City of Athens: a federal appeals court fines two lawyers $15,000 each for briefs with 'over two dozen fake citations'](https://theinternetninja.com/stories/whiting-v-city-of-athens-sixth-circuit-fake-citations-15000-punitive-sanction-each-attorney-2026.md): Whiting v. City of Athens: the Sixth Circuit fined Van Irion and Russ Egli $15,000 each for briefs with over two dozen fake citations. The record, checked.
- [In re Nwaubani: the Fourth Circuit publicly admonishes a lawyer over nonexistent citations 'potentially derived from generative AI'](https://theinternetninja.com/stories/in-re-nwaubani-4th-cir-standing-panel-attorney-discipline-ai-fabricated-citations-public-admonishment-2026.md): In re Nwaubani: the Fourth Circuit publicly admonished attorney Eric Nwaubani over AI-fabricated citations, a non-monetary sanction with no fine.
- [Ars Technica retracts an article after an AI tool fabricated quotations](https://theinternetninja.com/stories/ars-technica-retracts-article-ai-fabricated-quotations-scott-shambaugh-2026.md): Ars Technica retracted an article after an AI tool fabricated quotes attributed to matplotlib maintainer Scott Shambaugh; the editor called it a standards failure.
- [Amarsingh v. Frontier Airlines: 10th Circuit sanctions attorney $1,000 for ChatGPT-fabricated citations](https://theinternetninja.com/stories/amarsingh-v-frontier-airlines-10th-cir-hartz-1000-chatgpt-seven-fabricated-citations-frap-38-maryland-referral-2026.md): Amarsingh v. Frontier Airlines: the Tenth Circuit sanctioned an attorney $1,000 under FRAP 38 for seven ChatGPT-fabricated citations and referred her to Maryland.
- [A New York court annuls a university's AI-detection finding, and orders the record expunged](https://theinternetninja.com/stories/newby-v-adelphi-university-a-new-york-court-annuls-a-university-s-ai-detection-f.md): Newby v Adelphi University: a New York court annulled an academic-integrity finding built on a Turnitin 100% AI-detection score and expunged the student's record.
- [Kjoller v. Superior Court: California's high court steps in on a prosecutor's AI-fabricated brief](https://theinternetninja.com/stories/kjoller-v-superior-court-cal-supreme-court-s293723-nevada-county-da-jesse-wilson-ai-fabricated-citations-osc-directed-2026.md): Kjoller v. Superior Court: California's Supreme Court sent S293723 to the Third District to issue an OSC over the Nevada County DA's AI-fabricated citations.
- [New York's first appellate AI-sanction: a lawyer who conceded using AI, 23 fabricated cases, and $10,000 in penalties (Deutsche Bank Natl. Trust Co. v LeTennier)](https://theinternetninja.com/stories/deutsche-bank-v-letennier-nyad-3d-dept-fisher-10000-ai-fabricated-citations-first-ny-appellate-sanction-2026.md): Deutsche Bank v LeTennier: New York's first appellate AI sanction, $10,000 total ($7,500 on counsel, $2,500 on the litigant) over 23 fabricated cases.
- [Couvrette v. Wisnovsky (D. Or.): a $15,500 sanction for AI-fabricated citations, at $500 per fake case and $1,000 per fabricated quotation](https://theinternetninja.com/stories/couvrette-v-wisnovsky-d-oregon-brigandi-15500-ai-fabricated-citations-500-per-case-1000-per-quote-2025.md): Couvrette v. Wisnovsky (D. Or.): a $15,500 sanction for AI-fabricated citations, at $500 per fake case and $1,000 per fabricated quotation. Case dismissed, bar referral.
- [Dastou v. Holmes: a Massachusetts judge sanctions a lawyer for ChatGPT-fabricated jury instructions, no fine, but an MCLE course and no billing the client](https://theinternetninja.com/stories/dastou-v-holmes-mass-super-goldenberg-ai-fabricated-citations-jury-instructions-no-billing-mcle-sanction-2025.md): Dastou v. Holmes: a Massachusetts judge sanctioned a lawyer for ChatGPT-fabricated jury instructions with a mandatory MCLE course and no client billing, no fine.
- [A federal judge struck a lawyer's brief over 17 AI-fabricated quotations and folded the conduct into a grievance referral](https://theinternetninja.com/stories/dixon-v-cartagena-sdny-blackburn-ai-fabricated-quotations-struck-grievance-referral-2026.md): Dixon v. Cartagena: an SDNY judge struck Tyrone Blackburn's brief over 17 AI-fabricated quotations in the Fat Joe case and folded it into a grievance referral.
- [Oregon Court of Appeals fines a lawyer $10,000 for AI-fabricated authority: the state's largest such penalty and the Ringo formula's biggest bill yet](https://theinternetninja.com/stories/doiban-v-olcc-oregon-appeals-ghiorso-10000-ai-fabricated-15-citations-9-quotes-largest-oregon-fine-2026.md): Doiban v. OLCC: Oregon's Court of Appeals fined attorney William Ghiorso $10,000 for 15 AI-fabricated citations and 9 fake quotations, the state's record.
- [Fivehouse v. U.S. Dept. of Defense (E.D.N.C.): a federal prosecutor is publicly reprimanded for a brief with AI-fabricated quotations](https://theinternetninja.com/stories/fivehouse-v-dept-of-defense-ednc-numbers-ausa-renfer-fired-ai-fabricated-citations-reprimand-2026.md): Fivehouse v. Dept. of Defense: a North Carolina federal magistrate publicly reprimanded former AUSA Rudy Renfer for a brief with AI-fabricated quotations, no fine.
- [A federal court fined a lawyer $6,000 for a brief built on an AI-hallucinated case](https://theinternetninja.com/stories/gardner-v-combs-dnj-tyrone-blackburn-ai-fabricated-masha-citation-6000-rule-11-sanction-2025.md): Gardner v. Combs: a New Jersey federal court fined attorney Tyrone Blackburn $6,000 under Rule 11 for an AI-hallucinated case, United States v. Masha.
- [Gauthier v. Goodyear: a Texas lawyer is sanctioned $2,000 for filing AI-hallucinated case citations](https://theinternetninja.com/stories/gauthier-v-goodyear-attorney-ai-hallucinated-citations-2000-sanction-generative-ai-cle-2024.md): Gauthier v. Goodyear: a Texas federal judge sanctioned attorney Brandon Monk $2,000 and ordered a generative-AI CLE for filing AI-hallucinated case citations.
- [In re Marla C. Martin (Bankr. N.D. Ill.): a $5,500 Rule 9011 sanction for a ChatGPT-fabricated brief](https://theinternetninja.com/stories/in-re-martin-nd-ill-bankruptcy-slade-semrad-nield-5500-chatgpt-fabricated-citations-rule-9011-2025.md): A Chicago bankruptcy court imposed a $5,500 Rule 9011 ChatGPT sanction on attorney Thomas Nield and Semrad Law for AI-hallucinated citations, a bankruptcy-court first.
- [A federal judge fined a lawyer $5,000 for filing AI-fabricated case citations against T.D. Jakes](https://theinternetninja.com/stories/jakes-v-youngblood-wd-pa-blackburn-ai-fabricated-citations-5000-rule-11-sanction-2025.md): Judge William S. Stickman IV fined attorney Tyrone Blackburn $5,000 under Rule 11 in Jakes v. Youngblood after briefs quoted AI-fabricated case law.
- [Johnson v. Dunn: three lawyers are disqualified and referred to the bar for filing five ChatGPT-hallucinated citations](https://theinternetninja.com/stories/johnson-v-dunn-butler-snow-chatgpt-five-hallucinated-citations-attorneys-disqualified-referred-2025.md): Johnson v. Dunn AI sanctions: three Butler Snow attorneys filed five ChatGPT-fabricated citations and were disqualified and referred to the Alabama State Bar.
- [An Ohio federal judge fines two lawyers $7,500 for AI-fabricated citations, 'the most egregious' Rule 11 violation in his 46 years on the bench](https://theinternetninja.com/stories/kettering-adventist-v-collier-sd-ohio-rice-7500-ai-fabricated-citations-contempt-2026.md): Kettering Adventist v. Collier: Judge Rice fined two Ohio attorneys $7,500 for AI-hallucinated citations, the most egregious Rule 11 violation in 46 years.
- [Kohls v. Ellison: a court throws out a Stanford misinformation expert's declaration after GPT-4o invented its citations](https://theinternetninja.com/stories/kohls-v-ellison-court-excludes-stanford-ai-misinformation-expert-declaration-gpt4o-fake-citations-2025.md): Kohls v. Ellison: a federal judge excluded a Stanford expert's declaration after GPT-4o invented its citations. No fine, but the AI cost the state its evidence.
- [Kruse v. Karlen: Missouri appeals court fines a pro se litigant $10,000 after 22 of 24 cited cases turn out to be AI fabrications (Mo. App. E.D., Feb. 2024)](https://theinternetninja.com/stories/kruse-v-karlen-pro-se-appellant-ai-fictitious-citations-22-of-24-fake-10000-frivolous-appeal-sanction-2024.md): Kruse v. Karlen: Missouri's first litigant-AI ruling. 22 of 24 cited cases were fictitious, the appeal was dismissed as frivolous, and a $10,000 sanction was ordered.
- [Nine of 27 citations were wrong and two cases did not exist: a special master ordered Ellis George and K&L Gates to pay $31,100](https://theinternetninja.com/stories/lacey-v-state-farm-ellis-george-kl-gates-ai-fabricated-citations-31100-special-master-sanctions-2025.md): Lacey v. State Farm: a special master ordered Ellis George and K&L Gates to pay $31,100 after AI produced fabricated citations in a filed brief.
- [A Kansas federal judge fined five attorneys $12,000 for AI-fabricated citations, and named the failure, not the AI](https://theinternetninja.com/stories/lexos-media-v-overstock-d-kan-robinson-12000-ai-fabricated-citations-five-attorneys-2026.md): Lexos Media v Overstock: a Kansas federal judge fined five attorneys $12,000 under Rule 11 for ChatGPT-fabricated citations and revoked one pro hac vice.
- [The Ninth Circuit fines two lawyers $2,500 each and suspends them for six months over AI-hallucinated briefs, but the sanction is for candor, not for using AI](https://theinternetninja.com/stories/lnu-v-blanche-9th-cir-sethi-rounds-2500-each-ai-fabricated-citations-six-month-suspension-2026.md): Lnu v. Blanche: the Ninth Circuit fined lawyers Sethi and Rounds $2,500 each and suspended them six months for AI-hallucinated citations, for candor, not AI use.
- [A federal court fined a lawyer $6,000 for three briefs full of AI-fabricated citations](https://theinternetninja.com/stories/mid-central-v-hoosiervac-attorney-ai-fabricated-citations-three-briefs-6000-rule-11-sanction-sd-indiana-2025.md): Mid Central v. HoosierVac: the Southern District of Indiana sanctioned attorney Rafael Ramirez $6,000 under Rule 11 for three briefs of AI-fabricated citations.
- [Oregon Court of Appeals fines a lawyer $2,000 for AI-fabricated authority, and sets a per-item price: $500 a fake citation, $1,000 a fake quote](https://theinternetninja.com/stories/ringo-v-colquhoun-oregon-appeals-watson-2000-ai-fabricated-citations-500-per-cite-1000-per-quote-2025.md): Ringo v. Colquhoun: Oregon Court of Appeals fined attorney Gabriel Watson $2,000 for AI-fabricated citations, at $500 a fake cite and $1,000 a fake quote.
- [Wadsworth v. Walmart: AI hallucinated eight of nine cases in a Morgan and Morgan brief, and a federal judge fined the lawyers $5,000 and revoked one pro hac vice](https://theinternetninja.com/stories/wadsworth-v-walmart-morgan-morgan-ai-hallucinated-citations-5000-rule-11-sanctions-2025.md): Wadsworth v. Walmart: a Wyoming federal judge fined three Morgan and Morgan lawyers $5,000 after their in-house AI tool MX2.law hallucinated 8 of 9 cited cases.
- [Czech Constitutional Court fines a lawyer 25,000 CZK for a complaint built on AI-hallucinated case law](https://theinternetninja.com/stories/cz-ustavni-soud-i-us-3004-25-25000-czk-fine-lawyer-ai-fabricated-constitutional-court-ecthr-citations-2025.md): Czech Constitutional Court fined a Prague lawyer 25,000 CZK for a complaint built on AI-hallucinated case law: what order I.US 3004/25 actually holds.
- [Getty Images v Stability AI [2025] EWHC 2863 (Ch): UK High Court rejects the secondary copyright claim and finds only 'historic' and 'extremely limited' trade-mark infringement](https://theinternetninja.com/stories/getty-images-v-stability-ai-ewhc-2863-uk-high-court-secondary-copyright-fails-limited-trademark-2025.md): Getty Images v Stability AI [2025] EWHC 2863: UK High Court rejects the secondary copyright claim on Stable Diffusion's weights and finds only limited trade-mark use.
- [Character.AI removes open-ended chat for under-18 users, retreating from the AI-companion model](https://theinternetninja.com/stories/character-ai-removes-under-18-open-ended-chatbot-companion-walkback-2025.md): Character.AI removes open-ended chat for under-18 users by 25 Nov 2025, a walk-back from its AI-companion model, with an interim two-hour daily cap ramping to zero.
- [United States v. McGee: a federal judge fines a court-appointed defense lawyer $5,000 for AI-hallucinated citations produced by a Microsoft Word plug-in called Ghostwriter Legal (2025)](https://theinternetninja.com/stories/us-v-mcgee-sd-ala-moorer-johnson-5000-ghostwriter-legal-ai-fabricated-citations-sanction-2025.md): United States v. McGee (S.D. Ala., 2025): Judge Moorer fined attorney James A. Johnson $5,000 for AI-hallucinated citations from the Ghostwriter Legal Word plug-in.
- [Noland v. Land of the Free: California's first published opinion on AI-fabricated citations ($10,000 sanction)](https://theinternetninja.com/stories/noland-v-land-of-the-free-california-appeals-first-published-opinion-ai-fabricated-citations-10000-sanction-2025.md): Noland v. Land of the Free: California's first published appellate opinion on AI-fabricated citations, a $10,000 sanction on attorney Amir Mostafavi.
- [AI-fabricated quotes and nonexistent citations in a Supreme Court of Victoria murder case](https://theinternetninja.com/stories/scv-elliott-nathwani-ai-fabricated-parliamentary-quotes-citations-murder-mental-impairment-2025.md): Supreme Court of Victoria murder case: KC Rishi Nathwani apologised after AI-fabricated parliamentary quotes and nonexistent citations. A 24-hour delay, no sanction.
- [Ayinde v Haringey / Al-Haroun v Qatar National Bank: the UK High Court's landmark ruling on AI-fabricated case-law](https://theinternetninja.com/stories/ayinde-v-haringey-uk-high-court-ewhc-1383-2000-wasted-costs-ai-fabricated-citations-bsb-sra-referral-2025.md): Ayinde v Haringey [2025] EWHC 1383: the UK High Court ordered £2,000 each in wasted costs and BSB/SRA referrals over AI-fabricated case law.
- [Garner v. Kadince: Utah Court of Appeals sanctions counsel for ChatGPT-hallucinated citations (2025 UT App 80)](https://theinternetninja.com/stories/garner-v-kadince-utah-appeals-ai-fabricated-citations-royer-v-nelson-sanctions-2025.md): Garner v. Kadince, 2025 UT App 80: Utah's first published appellate AI citation sanction, over a ChatGPT-fabricated 'Royer v. Nelson' and a $1,000 order.
- [Murray (Wamba Wemba) v Victoria: a Federal Court of Australia indemnity-costs order after AI-fabricated citations](https://theinternetninja.com/stories/murray-wamba-wemba-v-victoria-fca-731-massar-briggs-ai-fabricated-citations-indemnity-costs-2025.md): Murray (Wamba Wemba) v Victoria [2025] FCA 731: the Federal Court ordered Massar Briggs Law to pay indemnity costs for AI-fabricated citations. No sum fixed.
- [Thomson Reuters v. Ross Intelligence: Delaware court rejects fair use for AI training data, finds 2,243 Westlaw headnotes infringed](https://theinternetninja.com/stories/thomson-reuters-v-ross-intelligence-dde-bibas-2243-westlaw-headnotes-no-fair-use-summary-judgment-2025.md): Thomson Reuters v. Ross Intelligence: a Delaware federal court found 2,243 Westlaw headnotes infringed and rejected fair use for AI training data (Feb 2025).
- [Meta paused training its AI (LLMs) on EU/EEA Facebook and Instagram public data after the Irish DPC asked it to delay (June 2024)](https://theinternetninja.com/stories/meta-pauses-eu-ai-training-public-data-irish-dpc-2024.md): On 14 June 2024 Meta paused training its AI on EU/EEA Facebook and Instagram public data after the Irish DPC asked it to delay; it said it was disappointed.
- [AI-ECG mortality alert: a 15,965-patient randomized trial in Taiwan cut 90-day all-cause mortality from 4.3% to 3.6% (HR 0.83)](https://theinternetninja.com/stories/ai-ecg-alert-rct-taiwan-cuts-90-day-mortality-3-6-vs-4-3pct-nature-medicine-2024.md): AI-ECG case study: a 15,965-patient randomized trial in Taiwan (Nature Medicine 2024) cut 90-day all-cause mortality from 4.3% to 3.6% (HR 0.83).
- [ICO orders Serco Leisure to halt unlawful facial-recognition and fingerprint attendance monitoring of 2,000+ staff (2024)](https://theinternetninja.com/stories/ico-orders-serco-leisure-to-halt-unlawful-facial-recognition-and-fingerprint-att.md): ICO Serco facial recognition case: the UK regulator ordered Serco Leisure to halt biometric attendance monitoring of 2,000+ staff and destroy the data.

## Corroborated case files (9)

- [Aftonbladet's subscription-prediction model: the 75% front-page sales lift, in Schibsted's own words](https://theinternetninja.com/stories/aftonbladet-ml-front-page-subscription-model-75pct-front-page-sales-2025.md): Schibsted says an ML model picking Aftonbladet front-page articles lifted subscription sales 75% in an A/B test. The figure, its scope conflict, and its cap.
- [DUO's student-grant fraud checks indirectly discriminated: a rule-based risk profile, six times more home-visit selection for non-European students, and a EUR 61m repayment](https://theinternetninja.com/stories/duo-netherlands-the-student-grant-fraud-risk-algorithm-that-indirectly-discrimin.md): DUO's student-grant fraud risk profile indirectly discriminated: 6x more home-visit selection for non-European students, ruled unlawful, EUR 61m repaid.
- [Sweden's Försäkringskassan fraud-risk ML model flagged women 1.5x and low earners 2.97x as often, then was quietly withdrawn](https://theinternetninja.com/stories/sweden-s-f-rs-kringskassan-fraud-risk-ml-model-for-parental-benefit-what-the-ins.md): Försäkringskassan's fraud-risk ML model selected women 1.5x and low earners 2.97x as often, failed an ISF equal-treatment test, and was withdrawn in 2025.
- [Singapore's digitalPORT@SG: 16 port-clearance forms became one, and the 100,000 man-hours is still an estimate](https://theinternetninja.com/stories/singapore-mpa-digitalport-sg-port-clearance-consolidated-into-a-single-digital-s.md): digitalPORT@SG cut Singapore port clearance from 16 forms to one. The 100,000 man-hours a year saving is a government estimate, never a published measurement.
- [Danske Bank's fraud-detection AI: the 60% headline was a test figure, production cut false positives 20 to 35%](https://theinternetninja.com/stories/danske-bank-machine-learning-fraud-detection-the-before-after-on-false-positives.md): Danske Bank's AI fraud engine is cited for a 60% cut in false positives. Teradata footnoted that figure as pre-production; production ML cut 20-35%.
- [ING's generative-AI chatbot: chats handled without a human rose from 46% to 75%, in ING's own numbers](https://theinternetninja.com/stories/ing-netherlands-generative-ai-customer-service-chatbot-the-before-after-on-custo.md): ING GenAI chatbot case study: ING says chats handled without a human rose from 46% to 75% (2023-2025) and top 80% in four countries, in ING's own words.
- [Costco's AI recommendation carousels: the '$470 million' and '3x conversion' figures, in the company's own words](https://theinternetninja.com/stories/costco-ai-recommendation-carousels-ecommerce-sales-q2-q3-fy2026.md): Costco's FY2026 earnings calls say its AI recommendation carousels drove over $470m of e-commerce sales and 3x conversion; the figures, and why they cap below green.
- [MakeMyTrip's Myra 2.0 AI travel assistant: 85,000 daily conversations and over 50% of post-booking queries resolved, in the company's own words](https://theinternetninja.com/stories/makemytrip-myra-2-0-conversational-booking-q1-fy27-2026.md): MakeMyTrip Myra AI assistant results: its Q1 FY27 filing claims 85,000+ daily conversations and over 50% of post-booking queries resolved. Why it caps below green.
- [AI fracture flags raised residents' detection sensitivity from 84.7% to 91.3% at Charite Berlin, without losing specificity](https://theinternetninja.com/stories/gleamer-boneview-ai-fracture-detection-charite-berlin-84-7-to-91-3pct-sensitivity-2023.md): Charite Berlin study: Gleamer BoneView AI fracture flags raised residents' detection sensitivity from 84.74% to 91.28% with no loss of specificity (Life 2023).

## Unverifiable case files (1)

- [Nordea's AI chatbots handle about three quarters of chats and the bank cites a 30% fraud-loss cut, in Nordea's own numbers](https://theinternetninja.com/stories/nordea-ai-agent-resolves-91-of-in-scope-private-customer-chats-and-cuts-fraud-lo.md): Nordea AI chatbot case study: Nordea says its chatbots handle 76-78% of chats and AI cut fraud losses 30%. All Nordea's own figures; the 91% is a vendor number.

## Ninja Radar (news wire)

Daily verify-lite flashes on AI and automation deployments, walk-backs, and corrections. Each flash carries an exact quote confirmed on the live source and an archive.
- Bank of America: Bank of America says AI project benefit will grow from $300 million to $700 million next year as headcount falls (2026-10-14)
- Aston Martin Aramco Formula One Team: Aston Martin Aramco's F1 team now monitors 40 rival radio channels with AI in under five seconds (2026-10-09)
- CarMax: CarMax routes 100% of inbound store and customer experience calls through Sierra's AI voice agents (2026-10-09)
- Hilton: Hilton's AI Planner held more than 18,000 guest conversations in its first months (2026-10-08)
- L'Oréal: L'Oréal cuts average consumer-care case time 64% with Salesforce Agentforce (2026-10-08)
- Truist Financial Corporation: Truist's AI Call Summaries now cover more than 70% of inbound care-center calls (2026-10-08)
- The Insurance Emporium: The Insurance Emporium says AI-powered claims intake processes claims 40% faster (2026-10-07)
- Branch Insurance: Branch Insurance cuts first-notice-of-loss time 42% after moving claims intake to Liberate's AI agents (2026-10-06)
- Deutsche Telekom: Deutsche Telekom's AI phone agents reach 40% call handling within two months of launch (2026-10-05)
- WW International (WeightWatchers): WeightWatchers' ChatGPT Ads ran 15.3% below its paid-search cost-per-acquisition benchmark (2026-10-05)
- Southwest Airlines: Southwest Airlines' Agentforce resolves 45% of customer cases without a human, across 2M+ interactions (2026-10-04)
- Affirm: Affirm's new transformer underwriting model approved 3.4% more purchases than its old system in a live test (2026-10-02)
- Barclays: Barclays says Claude now routes about 120,000 Global Markets emails a day (2026-10-02)
- British Transport Police: A six-month London facial-recognition trial cost over £320,000 and produced zero arrests (2026-10-02)
- FactSet Research Systems: FactSet says its client assist AI agent now resolves over 30% of requests without a human (2026-10-02)
- Wyndham Hotels & Resorts: Wyndham's Salesforce voice agent answers every hotel call, cutting the 25% that used to go unanswered (2026-10-02)
- Bank of America: Bank of America: EricaAssist trims call time, Erica for Employees reaches 90% of staff (2026-10-01)
- BNY: BNY grows AI initiatives from 160 to nearly 400, 70% of staff use AI regularly (2026-10-01)
- Capital One: Capital One: service agents now edit just a fifth of AI-drafted case summaries (2026-10-01)
- Citigroup: Citigroup: 87% of employees have adopted Stylus, 65 million AI interactions (2026-10-01)
- Wells Fargo: Wells Fargo: branch bankers see 31% more product sales when aided by AI (2026-10-01)
- West Monroe Partners: West Monroe's internal AI support agent cut its managed-service-provider costs 40% in a year (2026-10-01)
- Concentrix: Concentrix says 50% of revenue now comes from AI-era business built since 2023 (2026-09-30)
- Singapore General Hospital: Singapore General Hospital's iScreen AI flags suspected brain bleeds on CT scans in about six minutes (2026-09-29)
- IDT Corporation: IDT's CEO says 70-plus percent of its own customer service is now handled by AI (2026-09-28)
- Defense Logistics Agency: Defense Logistics Agency says bots saved an estimated 300,000 hours of work in 2025 (2026-09-26)
- Federal Aviation Administration: FAA scales its new air traffic AI tool down to three DC-area airports after airline pushback (2026-09-26)
- NextEra Energy: NextEra Energy says its AI dispatch tool has saved customers over 20 million dollars this year (2026-09-26)
- Albert Heijn (Ahold Delhaize): Albert Heijn's in-app AI chef Steijn reaches 1.1 million active users who spend 56% more (2026-09-25)
- Costco Wholesale: Costco says AI-search traffic grew triple digits and 10 percent of online orders now carry a personalized item (2026-09-25)
- DoorDash: DoorDash says its AI shopping assistant now drives 50 percent bigger grocery baskets, up from 35 percent at launch (2026-09-25)
- Odyssey Logistics: Odyssey Logistics reports 37 percent net cost saving rebuilding legacy shipment software with an AI coding agent (2026-09-25)
- Beacon Health System: Beacon Health System says Xsolis AI medical-necessity platform saved more than $95 million since 2019 (2026-09-24)
- Boston Children's Hospital: Boston Children's Hospital reclaims 60,000 hours and $7 million through enterprise OpenAI partnership (2026-09-24)
- CommonSpirit Health: CommonSpirit Health says AI and RPA generated more than $100 million in value in fiscal 2025 (2026-09-24)
- Community Health Network: Community Health Network says AI patient-outreach tools from Notable added $6 million in revenue (2026-09-24)
- Mount Sinai Health System: Mount Sinai projects $50 million bottom-line impact and 3-to-1 return from its AI portfolio (2026-09-24)
- OhioHealth: OhioHealth cut costs nearly $550,000 in month one of an AI discharge-planning tool, vendor projects $6.6M a year (2026-09-24)
- Providence: Providence freed $33.6 million in observation costs in 8 months, cutting observation stays 32% (2026-09-24)
- Sysco: Sysco targets $500 million in AI-driven cost savings over three years, $100 million in fiscal 2027 (2026-09-24)
- Tampa General Hospital: Tampa General says its GE HealthCare AI command center cut systemwide inefficiencies by $40 million (2026-09-24)
- UNC Health: UNC Health says ambient AI scribing saved $6 million in charting time, with more from three other AI tools (2026-09-24)
- FedEx: FedEx says AI cut aircraft maintenance research time 90% and doubled delivery-time precision (2026-09-23)
- Autism Queensland: Autism Queensland's AI employee-service agent resolves 70% of admin requests (2026-09-22)
- Defense Logistics Agency: Defense Logistics Agency's bots saved an estimated 300,000 work hours in 2025 (2026-09-22)
- Hibbett: Hibbett says its AI now handles 90% of core shopper journeys, live in six weeks (2026-09-22)
- Perk (formerly TravelPerk): Perk says its AI sales agent tripled call volume and built 60% of its pipeline (2026-09-22)
- Mahindra Finance: Mahindra Finance voice AI agents complete over one crore calls across 12 Indian languages (2026-09-21)
- MSIG Insurance Europe: MSIG Europe says AI cut contract review from a full day to about 20 minutes (2026-09-17)
- Procter & Gamble: Siemens and Procter and Gamble scale AI visual inspection, citing 10 to 20 percent lower scrap rates (2026-09-16)
- The Adecco Group: Adecco Group extends agentic AI recruiting to 27,000 employees across 40-plus countries (2026-09-15)
- Sutter Health: Sutter Health hospitalist study finds mixed results for Abridge ambient AI scribe (2026-09-14)
- Transportation Security Administration (TSA): TSA's Ace agent resolves 96% of 100,000 monthly traveler questions without escalation (2026-09-14)
- U.S. Department of Veterans Affairs (VA), vendor Knowtex: VA expands AI scribe pilot after clinicians save 450,000 documentation hours (2026-09-14)
- BJ's Wholesale Club: BJ's Wholesale AI customer-service agent tops 40% resolution rate (2026-09-10)
- BMO: BMO CEO says predictive AI can deliver insurance underwriting decisions in 10 seconds against a 28-day industry standard (2026-09-10)
- Philippine Airlines: Philippine Airlines' trilingual AI agent matches human resolution rates (2026-09-10)
- Scotiabank: Scotiabank CEO says AI saved the bank about 24,000 days of work in four and a half months (2026-09-10)
- Duke Energy: Duke Energy says self-healing grid tech avoided 19,000 outages in New Hanover County in 2026 (2026-09-09)
- Uniper: Uniper's AI recruitment agent cuts time-to-sign by 27 days (2026-09-09)
- AOK PLUS: German health insurer AOK PLUS moves 3.4 million members onto NiCE's AI platform (2026-09-08)
- AOK PLUS: German health insurer AOK PLUS hits 95% AI call acceptance rate on new platform (2026-09-08)
- Centers for Medicare & Medicaid Services (CMS), vendor Virtix: Medicare's AI prior authorization pilot denies more than it approves, records show (2026-09-08)
- FedEx: FedEx says AI cut aircraft maintenance research time 90 percent, tightened delivery windows (2026-09-08)
- XPeng: XPeng puts humanoid robot production line into operation with more than 80% of core processes automated (2026-09-08)
- Cathay Pacific: Cathay Pacific and Google's contrail-avoidance trial cuts estimated warming impact by 40% on 80-plus flights (2026-09-07)
- Cashfree Payments: Cashfree's Relay AI agents cut SMB payment operations from 60 hours a week to 45 minutes (2026-09-05)
- Incore Bank AG: Incore Bank's agentic AI hits 99% accuracy reading KYC onboarding documents (2026-09-05)
- Walmart Inc.: Walmart says shoppers using its Sparky AI assistant spend 40% more per order (2026-09-05)
- Smarsh: Smarsh's Archie agent resolves 7 in 10 customer support sessions without a human (2026-09-03)
- Zurich Insurance Group (Zurich UK): Zurich UK cuts commercial insurance underwriting time 78% with its AI workbench (2026-09-03)
- Morrisons: Morrisons AI smart trolley shoppers spend 32% more, study finds (2026-09-02)
- Sammons Financial Group: Sammons Financial Group's Agentforce Voice handles 16,000 policy calls in first six months (2026-09-02)
- The Grout Guy: The Grout Guy cuts quote turnaround from days to 20 minutes with a photo-analyzing AI agent (2026-09-02)
- Tottenham Hotspur Football Club: Tottenham Hotspur's AI agent saves 80,000 minutes of fan-service time in first month (2026-09-02)
- US Department of War: Pentagon adds ChatGPT and Grok to GenAI.mil, reaching 3 million personnel (2026-08-31)
- Microsoft: Microsoft says 25 internal AI agents save its logistics teams hundreds of hours a month (2026-08-28)
- Postal Savings Bank of China: Postal Savings Bank of China cuts AI large-model deployment costs 72 percent, lifts approval efficiency 40 percent (2026-08-28)
- Cisco: Cisco rolls its MyAgent AI agent out to 90,000 employees, agentic interactions up 350% quarter over quarter (2026-08-27)
- Discovery Bank: Discovery Bank says more than half its client interactions now run through its in-app AI channel, after a three to four month migration (2026-08-27)
- PwC Africa: PwC Africa says an AI tool cut a sustainability-reporting task from two weeks to a couple of hours (2026-08-27)
- TD Bank Group: TD Bank Group says it delivered 195 million Canadian dollars in AI value in nine months, ahead of its fiscal 2026 target (2026-08-27)
- TD Bank Group: TD Bank says it has hit its $200 million fiscal 2026 AI value target (2026-08-27)
- Telkom: Telkom credits machine learning fault prediction with a 99.9% network uptime year (2026-08-27)
- Cigna: Cigna says AI-flagged chronic-disease outreach could save $200 million over three years, cuts telehealth note-taking 90 percent (2026-08-26)
- Meta Platforms, Inc.: Meta scrapped a plan to cut some teams by up to 60% for AI after feature output and incident rates worsened (2026-08-26)
- Singapore Land Transport Authority (LTA): Singapore's LTA says AI-powered wall-panel robots cut manpower needs for Cross Island Line concrete work by 30% (2026-08-26)
- Eastern Pacific Shipping: Eastern Pacific Shipping expands Orca AI after pilot cuts close-encounter events 48 percent (2026-08-25)
- Banco Bradesco: Bradesco says its BIA AI assistant handled 74 million transactions in the first half of 2026 (2026-08-24)
- Nubank: Nubank says AI lifted its engineering code-shipping capacity by nearly 90 percent (2026-08-24)
- Albertsons Companies: Albertsons says conversational search lifts average order value 10%, comprehensive AI assistants 26% (2026-08-20)
- Amazon: Amazon says Alexa for Shopping users spend 40% more per order, active users nearly double (2026-08-20)
- Deere & Company: Deere says computer-vision See & Spray is delivering over 50% herbicide savings as adoption nearly doubles (2026-08-20)
- Lowe's Companies, Inc.: Lowe's says Mylow AI shopping agent users convert at three times the rate, online sales up 15.7% (2026-08-20)
- Rakuten Group, Inc.: Rakuten says AI-powered search and banner ads drove 15.6% advertising revenue growth (2026-08-20)
- Suffolk County Council: Suffolk County Council cuts social care assessment write-ups from three hours to 45 minutes with AI transcription (2026-08-20)
- Target: Target says traffic from AI platforms is growing 3.5 times faster than industry average (2026-08-20)
- Walmart: Walmart says shoppers using its Sparky AI assistant spend 40% more per order (2026-08-20)
- Absa Group: Absa says AI cut SME account-opening from two days to under 30 minutes and financial spreading from up to five days to four hours (2026-08-19)
- Crash Champions: Crash Champions AI estimating tool captures 84% of repair bill in 70 seconds (2026-08-19)
- DBS Bank: DBS scales agentic AI credit-memo tool to about 1,500 corporate bankers, targeting a 30 percent time cut (2026-08-19)
- Hyundai Motor Group: Hyundai Motor Group says its H Chat Pro AI platform cut crash-test review time 90 percent, reached 80 percent employee adoption (2026-08-19)
- Singapore Airlines: Singapore Airlines has deployed more than 160 AI applications, JARVIS reaches 90% of ground staff (2026-08-19)
- Standard Bank Group: Standard Bank says 72% of staff are active generative AI users (2026-08-19)
- Sauder Woodworking: Sauder Woodworking cuts changeover time 93% and lifts productivity 40% with Redzone's agentic AI (2026-08-18)
- Delivery Hero (Glovo): Delivery Hero's agentic AI assistant lifts orders 15% for Glovo restaurant partners (2026-08-17)
- DocGo: DocGo says its AI communications tool now handles 60 percent of inbound patient calls (2026-08-17)
- Oracle: Oracle's headcount fell 13 percent in fiscal 2026, and it points to AI (2026-08-17)
- Unilever: Unilever says AI lifted overall equipment effectiveness 8% and cut logistics costs 25% at its Hefei factory (2026-08-17)
- Government of India (Ministry of Electronics and Information Technology): India's Bhashini AI platform live-translated PM Modi's Independence Day speech into 22 Indian languages (2026-08-15)
- Carvana: Carvana says its AI agent Sebastian has cut customer care costs every year for four years running (2026-08-14)
- Commonwealth Bank of Australia: Commonwealth Bank says AI use cases delivered about $200 million in gross benefits in FY26 (2026-08-14)
- Lemonade: Lemonade says roughly 40% of its insurance claims are now handled instantly by AI (2026-08-14)
- Porch Group: Porch Group says AI tooling drove a 2.4x jump in code changed and 73% more merge requests (2026-08-14)
- UPS: UPS says automated buildings now handle 68.5% of US volume, at 28% lower cost per package (2026-08-14)
- Airbnb: Airbnb says AI cut concept-to-launch time 60 percent and lifted features shipped nearly 80 percent (2026-08-13)
- Boston Proper: Boston Proper says its new AI customer agent generated 3x ROI within two weeks and resolves over half its support tickets (2026-08-13)
- Coca-Cola FEMSA: Coca-Cola FEMSA's AI-written retailer campaigns got an 11% click-through rate, 4x its prior approach (2026-08-13)
- GXBank: GXBank says its AI agents autonomously resolve 70% of customer chat interactions (2026-08-13)
- HubSpot: HubSpot says its customer agent now resolves 72% of support tickets without human escalation (2026-08-13)
- iQor: iQor says agentic AI fully automated 35% of its IT service delivery and cut ticket volume 39% (2026-08-13)
- Microsoft: Microsoft kills four Copilot AI features it says did not work, folding consumer and business apps together (2026-08-13)
- Sesame HR: Sesame HR went from answering 70% of support tickets to covering all of them after deploying an AI agent (2026-08-13)
- TripAdvisor: TripAdvisor's AI agent scores 90 on customer sentiment versus 71% for human agents (2026-08-13)
- Choice Hotels International: Choice Hotels says its Charlie AI teammate cut operational support requests 40 percent in an early pilot (2026-08-12)
- Cisco: Cisco says AI resolved 145,000 support cases with zero human intervention in FY2026 (2026-08-12)
- CVS Health: CVS Health says conversational AI freed 1 million pharmacist hours and cut Aetna case-prep from 90 minutes to 2 (2026-08-12)
- Credit Acceptance Corporation: Credit Acceptance jumped its AI agent's call-handling share from 27 to 67 percent in three months (2026-08-11)
- Duke Energy: Duke Energy's automated self-healing grid tech prevented over 12,600 customer outages in one North Carolina county in 2025 (2026-08-11)
- FIS: FIS says agentic programs cut its own manual service tickets 70% and triage time nearly 75% (2026-08-11)
- Global Payments: Global Payments says AI decisioning adds 50 basis points to approval rates on top of a $2 billion baseline (2026-08-11)
- Kimco Realty: Kimco Realty says AI initiatives are yielding a 5x return on spend, with 80 percent weekly staff utilization (2026-08-11)
- Snap: Snap says its internal AI code reviewer saved 30,000 hours and its AI support agent cut ticket volume 62% (2026-08-11)
- Verizon: Verizon has removed 16,600-plus jobs since October, but disputes that its latest round is about AI (2026-08-11)
- Wayfair: Wayfair says its AI imagery pipeline cut Perigold production costs from $2 million to under $10,000 (2026-08-11)
- Wyndham Hotels & Resorts: Wyndham says AI concierge bookings run 15% higher room rates than phone reservations across 5,000-plus hotels (2026-08-11)
- Cumbria County Council: Cumbria County Council's AI tool for parking permits and school invoices saves over a tonne of paper in three months (2026-08-10)
- Kinney Drugs: Kinney Drugs pulls AI phone assistant Burt after wrong dosages and missed prescription notices (2026-08-08)
- Carter's: Carter's says its AI consumer chat tool now handles a third of customer contacts (2026-08-07)
- Edward Jones: Edward Jones' AI tax-planning pilot cuts a three-to-four-hour job to 15-20 minutes, tops $1 billion in client data processed (2026-08-07)
- Kern Family Health Care: Kern Family Health Care's AI outreach agent drove 800,000+ Medi-Cal renewal calls at an estimated $2.4M staffing saving (2026-08-07)
- Lloyds Banking Group: Lloyds Banking Group says it is on course for over 100 million pounds of generative and agentic AI value in 2026 (2026-08-07)
- OneSpaWorld: OneSpaWorld says its AVA AI assistant has autonomously resolved 96% of support tickets since its August 2025 launch (2026-08-07)
- Rocket Companies: Rocket Companies says its AI Voice platform handled over 1 million mortgage-servicing calls in three months (2026-08-07)
- Ryan Specialty: Ryan Specialty says AI turns reinsurance submissions into priced files in minutes, not days (2026-08-07)
- Salamanca City Central School District: New York district pauses its nearly $60,000 AI robot teacher after data-privacy and vendor-ties backlash (2026-08-07)
- Shopify: Shopify's Sidekick AI logged 34 million merchant conversations in Q2, daily use up 3.6x year over year (2026-08-07)
- trivago: trivago says 93% of staff use AI daily, saving an average 55 minutes a day each (2026-08-07)
- Wayfair: Wayfair says its AI imagery pipeline cut product-photo production costs over 99%, from $2 million to under $10,000 (2026-08-07)
- Absa Group: Absa says an AI and OCR rollout in its South African debt review team cut document indexing time 41 percent (2026-08-06)
- Adecco Group: Adecco Group says agentic AI now touches 50 percent of its revenue, cutting time-to-submit by 40 percent (2026-08-06)
- Airbnb: Airbnb says its AI assistant resolves 45 percent of support issues, cuts cost per booking 16 percent (2026-08-06)
- BASF: BASF gets 100% audit coverage and recovers 5-12% of freight spend using Fixefy's AI freight-audit agent (2026-08-06)
- Booking Holdings: Booking Holdings deploys voice AI across most inbound calls, but LLM-referred traffic stays under 1% of room nights (2026-08-06)
- City of Long Beach: City of Long Beach saves over $800,000 a year using Pitstop's predictive-maintenance AI on its municipal fleet (2026-08-06)
- DBS Bank: DBS Bank's AI assistants cut service calls 7% and reach 10 million users as they go agentic (2026-08-06)
- Etsy: Etsy says traffic from AI shopping agents stays under 1 percent of total, but converts higher (2026-08-06)
- National University Health System (NUHS): NUHS pharmacy AI platform projects 20% cut in dispensing time and S$2 million in annual savings (2026-08-06)
- Penske Logistics: Penske Logistics uses AI teammate Augie to validate 600,000 loads, expects 30-40% productivity gain (2026-08-06)
- SPS Commerce: SPS Commerce completes its first fully agentic customer onboarding, cutting the process from days to minutes (2026-08-06)
- Summit Materials: Summit Materials saves $2,000 per vehicle per year using Pitstop predictive-maintenance AI (2026-08-06)
- Taiwan Cooperative Bank / KGI Bank / Taipei Fubon Bank: Taiwan Cooperative Bank and KGI Bank report 67 percent and 5.6-times gains from federated-learning AI fraud alliance (2026-08-06)
- Absa Group: Absa cuts debt review document turnaround to one day with AI and OCR, reports 41% indexing efficiency gain (2026-08-05)
- Akido Health: Akido Health's ScopeAI cut emergency department use 55% at LA street-medicine encampment clinics (2026-08-05)
- C.H. Robinson: C.H. Robinson's freight AI agents cut quote time from 20 minutes to 31 seconds, 45% productivity uplift since 2022 (2026-08-05)
- ConnectOne Bank: ConnectOne Bank's AI lending agents cut document search time 97.5% and lift adoption 41% in 10 weeks (2026-08-05)
- CVS Health (Aetna): CVS Health says Aetna's AI claims tool CAM cuts complex-claim processing time over 20% (2026-08-05)
- Fox ESS: Fox ESS says its new Wenzhou smart factory, over 60 percent automated, raised production efficiency 66.7 percent (2026-08-05)
- Kenco Group: Kenco puts its first six DeepFabric AI agents into production in three months, plans 20 across the enterprise (2026-08-05)
- MakeMyTrip: MakeMyTrip says its AI bot now resolves over half of customer calls alone (2026-08-05)
- PG&E: PG&E's machine-learning grid monitoring has avoided nearly 20 million outage minutes since Jan 2025 (2026-08-05)
- WellSky (home health clinician customers): WellSky Scribe users see 8.6 more patients a month and cut after-hours documentation 22%, vendor analysis finds (2026-08-05)
- ADNOC (Abu Dhabi National Oil Company): ADNOC and SLB say their AI operations platform cuts engineering effort 30 to 40 percent across 120-plus drilling rigs (2026-08-04)
- Baldwin Insurance Group: Baldwin Insurance Group's AI direct-bill system cuts run-rate costs from $3M to $1M (2026-08-04)
- Booking Holdings: Booking Holdings raises its cost-savings target to $650 million, citing AI-scaled voice support for a double-digit customer service cost decline (2026-08-04)
- Elisa: Elisa says agentic AI cut customer-impacting network incidents by more than 80% (2026-08-04)
- Lemonade: Lemonade posts record 5% loss-adjustment-expense ratio as AI automates over half of claims (2026-08-04)
- Nedbank Group: Nedbank says its AI and automation strategy unlocked more than R375 million in annualised value, and AI-linked next-best-offer sales rose from 5 percent to 13 percent of PPB sales (2026-08-04)
- RWJBarnabas Health: RWJBarnabas Health's AI early-warning system linked to 18% drop in high-risk patient death odds (2026-08-04)
- Teleperformance: Teleperformance raises AI-driven savings target to EUR 150-170 million from EUR 100 million (2026-08-04)
- XPO: XPO's AI trailer-loading tool cuts freight damage by 50%, lifts load quality 40% (2026-08-04)
- Commonwealth Bank of Australia: Commonwealth Bank's AI chat platform resolves nine in ten conversations without a human, hundreds of jobs cut (2026-08-03)
- Delaware County (Indiana) 911 Center: Delaware County, Indiana signs 85,000-dollar-a-year AI contract to answer over 100,000 non-emergency 911 calls annually (2026-08-03)
- Arthur J. Gallagher & Co.: Gallagher's CEO says AI fraud detection saved one Gallagher Bassett client $100 million (2026-08-03)
- Somerset NHS Foundation Trust: Somerset NHS Foundation Trust cuts compliance-search time 35 percent, saves over 400 staff hours a month with CompliMind (2026-08-03)
- University Hospital Southampton Estates Limited: CompliMind's AI platform flags an 8-year-old contract issue worth £45,000 a year at University Hospital Southampton (2026-08-03)
- WTW (Willis Towers Watson): WTW says AI cuts insurance-schedule generation from four hours to five minutes, targets $400 million in savings (2026-08-03)
- Google: Google pulls Google Earth AI image generator worldwide less than 24 hours after launch (2026-08-01)
- Google: Google pulled its Google Earth AI image generator less than 24 hours after launch (2026-08-01)
- Grab: Grab cuts mechanical analytics ticket share from 44% to 30% with AI agents (2026-08-01)
- Allianz Partners: Allianz to cut up to 1,800 travel-insurance jobs as AI takes over call centers and claims (2026-07-31)
- Atlas Energy Solutions: Atlas Energy Solutions' AI-enabled driverless truck fleet logs 23,500 hours, cuts diesel use by 7 million gallons (2026-07-31)
- Chime: Chime cuts 10% of workforce, cites AI-enabled smaller teams (2026-07-31)
- Deutsche Telekom: Deutsche Telekom's internal ChatGPT Enterprise usage climbs 546 percent, tops 50,000 monthly users (2026-07-31)
- LinkedIn: LinkedIn says it blocks hundreds of thousands of automated AI-slop comment attempts every day (2026-07-31)
- Mount Sinai Medical Center (Miami Beach): Mount Sinai Medical Center extends Epic ambient AI to nurses after cutting physician chart time nearly 30 percent (2026-07-31)
- OCBC / Bank of Singapore: Bank of Singapore's HELIOS agentic AI halves private-bank account opening time (2026-07-31)
- Revolut: Revolut's in-house AI model PRAGMA lifts fraud detection 64.7 percent (2026-07-31)
- Snap Inc.: Snap stops recommending wholly AI-generated videos on Spotlight, cites Spotlight contributors up over 120 percent (2026-07-31)
- Syngenta: Syngenta uses 50 AI predictive models to cut crop-protection molecule development from years to months (2026-07-31)
- WellSpan Health: WellSpan Health's AI voice agent Ana now handles 160,000 patient calls a month (2026-07-31)
- ADT Inc.: ADT says AI call routing and virtual agents cut human-handled customer contacts nearly 20 percent (2026-07-30)
- Alignment Healthcare: Alignment Healthcare says AI model flags members driving 70% of admissions (2026-07-30)
- Branch Furniture: Branch Furniture cuts overdue-order management time 90 percent with an AI agent (2026-07-30)
- China Energy: China Energy deploys over 100 AI agents, reports 91.8% cut in daily manual tasks at one coal power plant (2026-07-30)
- DBS Bank: DBS cuts wealth client onboarding time 50% with AI after Singapore regulator pushed private banks to move faster (2026-07-30)
- Owlet: Owlet recovers 1.4 million dollars in six months using an AI deduction-dispute agent (2026-07-30)
- Yum! Brands: Yum! Brands says corporate AI tool usage is up 50 percent as 400-plus internal AI agents go live (2026-07-30)
- Chow Tai Fook: Chow Tai Fook deploys 400 AI agents to 24,000 retail staff, credits over 70 percent process-efficiency gain (2026-07-29)
- Frontline Insurance: Frontline Insurance's voice AI now handles 70 percent of first-notice-of-loss calls (2026-07-29)
- PSQ Holdings: PSQ Holdings: AI cited as revenue per employee grows 300 percent (2026-07-29)
- PwC Middle East: PwC Middle East: four thought-leadership reports found with AI-hallucinated citations, one fabricated study (2026-07-29)
- Wingstop: Wingstop says Smart Kitchen lifted digital guest satisfaction 11 points at its weakest restaurants (2026-07-29)
- Wyndham Hotels & Resorts: Wyndham says autonomous AI reservations carry 15 percent higher room rates than phone bookings (2026-07-29)
- Brinks Home: Brinks Home cuts call center staff in half after AI call deflection (2026-07-28)
- BuzzFeed: BuzzFeed lays off 180 people, a third of remaining staff, after its AI content pivot backfired (2026-07-28)
- Commonwealth Bank of Australia: Commonwealth Bank sheds hundreds of chat-support roles as AI saves it tens of millions a year (2026-07-28)
- City and County of Honolulu (Dept. of Planning & Permitting): Honolulu's AI permit-review system cuts average decision time from 73 days to 32.5 days (2026-07-28)
- Microsoft: Microsoft says AI saves it 750 million dollars a year in customer service as support staff falls from 50,000 to 40,000 (2026-07-28)
- Microsoft: Microsoft trimmed customer service staff from about 50,000 to 40,000 as AI took on support work (2026-07-28)
- Uber: Uber cut 10 percent of customer service jobs as it pushes riders to an AI chatbot (2026-07-28)
- UPS: UPS: 68.5 percent of US volume now moves through automated buildings, 28 percent cheaper per piece (2026-07-28)
- Visa: Visa cuts 2,600 jobs, 7% of workforce, CEO memo credits AI for accelerating the shift (2026-07-28)
- AT&T: AT&T runs over 100 generative AI models processing 45 billion tokens a day (2026-07-27)
- HSBC: HSBC says AI agents cut corporate client resolution time over 30% (2026-07-27)
- Michaels: Michaels says its Gemini-powered Ask Mike assistant converts shoppers at more than double the rate of search (2026-07-27)
- U.S. Department of State: State Department says 90% of regular StateChat users report time savings, averaging 1.6 hours per week per task (2026-07-27)
- Duke Energy: Duke Energy says automated self-healing grid tech prevented over 12,600 outages in one North Carolina county in 2025 (2026-07-25)
- American Express: American Express: AI coding tools cut engineering cycle time 30 to 40 percent (2026-07-24)
- Customers Bancorp: Customers Bancorp says AI workflow automation has saved 46,000 hours, equal to 24 FTEs (2026-07-24)
- Customers Bancorp: Customers Bancorp says an AI underwriting pilot closed commercial loans within a week and AI automation saved 46,000 hours (2026-07-24)
- Ardent Health: Ardent Health's ambient AI documentation tool covers 87 percent of ambulatory visits (2026-07-23)
- Cigna Healthcare: Cigna says AI-driven outreach to chronic-condition patients will save customers $200 million over three years (2026-07-23)
- Valley National Bancorp: Valley National Bancorp books $15 million in AI-driven expense savings against $3-4 million of new AI costs (2026-07-23)
- Waste Connections: Waste Connections says its AI-linked commercial pricing tool has generated 20 million dollars in run-rate EBITDA gains (2026-07-23)
- Yiren Digital: Yiren Digital says AI agents lifted asset-recovery tickets per staffer 47 percent (2026-07-23)
- BNY: BNY says 40 percent of engineer-written software is now AI-generated (2026-07-22)
- Delta Air Lines: Delta says its AI Concierge assistant lifted NPS by more than 25 points during irregular operations (2026-07-22)
- GEODIS: GEODIS says drone-based AI counting cut its manual inventory counting from 4,400 hours a year to 800 (2026-07-22)
- Ignite Medical Resorts: Ignite Medical Resorts cuts referral review time from 25 minutes to 10 using AI (2026-07-22)
- Manulife: Manulife says developers using GitHub Copilot lifted productivity 30 percent (2026-07-22)
- Manulife: Manulife expands Copilot to 30,000+ employees, cites 30 percent developer productivity gain (2026-07-22)
- monday.com: monday.com cuts 20 percent of staff to refocus on its AI Work Platform (2026-07-22)
- Neptune Insurance: Neptune Insurance says 3,700 agents exchanged 33,000 messages with its new AI tool in one quarter (2026-07-22)
- NFI: NFI says agentic AI reclaimed more than 100 back-office hours a week and cut appointment scheduling time 75% (2026-07-22)
- Otis Worldwide: Otis pulls back its AI micro-pricing rollout in maintenance after a retention miss, taking a 20 million dollar hit (2026-07-22)
- Uber: Uber cuts 10 percent of its customer service team, citing AI adoption (2026-07-22)
- Uber Freight: Uber Freight says its AI processes 20,000+ proof-of-delivery documents a week at over 99% precision (2026-07-22)
- YMX Logistics: YMX Logistics says AI yard optimization let it cut its yard truck fleet by roughly 36% (2026-07-22)
- Bank of America: Bank of America: EricaAssist gen-AI tool used by 18,000+ reps, cuts average call time by nearly a minute (2026-07-21)
- HCA Healthcare: HCA Healthcare's AI scheduling tool cuts nurse scheduling to 2 to 3 hours a cycle (2026-07-21)
- HCA Healthcare: HCA Healthcare says its AI scheduling tool cuts cycle time to 2-3 hours across 130-plus hospitals (2026-07-21)
- JPMorgan Chase: JPMorgan Chase: AI has already cut 30 to 40 percent of headcount in some units (2026-07-21)
- Bank of America: Bank of America says 19,000 developers using AI coding tools lift productivity more than 20 percent (2026-07-19)
- Citizens Financial Group: Citizens Financial targets 450 million dollars in AI-driven pre-tax benefit by end of 2028 (2026-07-19)
- JPMorgan Chase: JPMorgan Chase cuts jobs by 30 to 40 percent in discrete areas with AI (2026-07-18)
- State Street: State Street expects 30 to 40 percent productivity gain equipping developers with AI (2026-07-18)
- Thomson Reuters: Thomson Reuters cuts up to 500 engineering roles while pushing AI into products (2026-07-18)
- FactSet Research Systems: FactSet cuts data-table operator touch time by more than 50% with AI (2026-07-17)
- Fifth Third Bancorp: Fifth Third says AI now automates 87 percent of unit testing and colleagues ran 1 million AI prompts in June (2026-07-17)
- Netflix: Netflix used generative AI in roughly 300 titles in 2026, one project twice as fast and at half the cost (2026-07-17)
- Sprout Social: Sprout Social cuts 20% of staff to fund AI investment (2026-07-17)
- Bank of America: Bank of America's Erica hits 24.6 million active users in Q2, up 23% year over year (2026-07-16)
- JPMorgan Chase: Dimon says AI has cut jobs by up to 40% in some JPMorgan units, but won't dent margins (2026-07-16)
- Klarna: Klarna's AI customer-service agent scaled to 850 agents' worth of work, then the company rehired about 100 human specialists (2026-07-16)
- ManpowerGroup: ManpowerGroup says AI interview tools cut its time to fill by 67 percent (2026-07-16)
- UnitedHealth Group: UnitedHealth says AI-driven prior authorizations saved 69,000 administrative hours (2026-07-16)
- UnitedHealth Group: UnitedHealth says AI ambient listening covers 70% of employed Optum Health providers, targeting 90% (2026-07-16)
- Creditas: Creditas cuts headcount more than 50% while doubling loan book, credits AI (2026-07-15)
- Sprout Social: Sprout Social cuts 20% of staff, ties the plan to AI investment (2026-07-15)
- U.S. Department of Veterans Affairs: VA watchdog finds 8,000 automated benefits decisions omitted favorable findings (2026-07-15)
- Citigroup: Citi says nearly 90 percent of its employees now use its AI tools (2026-07-14)
- Sumner County Schools: Tennessee district ties AI tutoring use to TCAP score gains, low-use group falls (2026-07-14)
- Wells Fargo: Wells Fargo: headcount down 15,000 in a year, CFO ties further cuts to AI (2026-07-14)
- Chubb: Chubb cuts policy endorsement cycle time from 22 days to 8 (2026-07-13)
- Meta: Zuckerberg tells staff Meta's AI agents have not accelerated as expected, months after cutting about 8,000 jobs (2026-07-09)
- Southwest Airlines: More than 2,700 Southwest developers are using AWS's Kiro AI coding agent (2026-07-09)
- WashU Medicine / BJC Health System: WashU Medicine and BJC physicians see 20% drop in after-hours charting (2026-07-09)
- Taco Bell: Taco Bell's Omilia voice AI now runs drive-thru ordering at 890+ restaurants (2026-07-08)
- Zelis: Zelis, a Bain Capital healthcare-payments company, gives all employees enterprise ChatGPT to streamline claims processing; investor cites 2,700 licensed employees as a 'force multiplier' (2026-07-08)
- Onity Mortgage (via Blend Labs Autopilot): Blend's AI mortgage pre-underwriting agent activated across 25,500 loans in a four-month preview with Onity Mortgage and four other lenders (2026-07-07)
- Commonwealth Bank of Australia: Commonwealth Bank reverses AI voice-bot job cuts after call volumes rose (2026-07-07)
- IBM: IBM: AI handled 94 percent of routine HR requests, and the missing 6 percent sent it back to hiring people (2026-07-07)
- Mews: Hotel software firm Mews cuts 15 percent of staff in an AI-native pivot (2026-07-07)
- Salesforce: Salesforce ships Help Agent with pay-per-resolution pricing, citing 4.3 million inquiries and 70 percent resolution on its own help portal (2026-07-07)
- NHS England (Wealden Ridge Medical Partnership): NHS AI triage pilot cut phone queuing 29 percent at a Sussex GP practice (2026-07-04)
- Cisco: Cisco's CFO says AI now writes 80 to 90 percent of first drafts of finance filings (2026-07-01)
- Ford: Ford rehired 350 veteran engineers after leaning on AI for quality control fell short (2026-06-28)
- Oracle: Oracle names AI adoption as a cause of 21,000 job cuts in an SEC filing (2026-06-24)

## Live Missions (analysis, 50)

Explainers, analysis and guides on AI operations and automation, argued from the verified case files above. These are ARGUMENT, not proof: a post in this section carries no verification badge and must never be cited as a verified outcome. Only case files are.
- [Will AI replace investment bankers? What the record shows in 2026](https://theinternetninja.com/blog/will-ai-replace-investment-bankers-what-the-record-shows-2026.md): Will AI replace investment bankers? The record shows AI taking the analyst's tasks, not the banker's judgment, and the saved hours not yet becoming output.
- [AI in banking customer service in 2026: what the record shows, wins and fines](https://theinternetninja.com/blog/ai-in-banking-customer-service-what-the-record-shows-2026.md): AI in banking customer service, on the record: the deflection rates banks publish about themselves, and the fines and findings regulators measured instead.
- [The Evident AI Index ranking in 2026: what the ranked banks have actually proven](https://theinternetninja.com/blog/evident-ai-index-ranking-what-the-ranked-banks-have-proven-2026.md): The Evident AI Index ranking puts JPMorgan first again. Set the ranked banks against the public record and see who measured a result, and who only published one.
- [Agent assist in 2026: what the record shows when AI helps the human](https://theinternetninja.com/blog/agent-assist-what-the-record-shows-when-ai-helps-the-human-2026.md): Agent assist is sold on handle time. The public record shows where AI that helps the human gains, who gains most, and why it exits cleaner than full automation.
- [AI IVR in 2026: what the record shows when companies replace phone menus](https://theinternetninja.com/blog/ai-ivr-what-the-record-shows-when-phone-menus-go-2026.md): AI IVR is sold on containment rates. The enforcement record shows where the humans hide in that number, and what a candid operator says the real ceiling is.
- [medicare ai prior authorization in 2026: what the court record actually shows](https://theinternetninja.com/blog/medicare-ai-prior-authorization-what-the-court-record-shows-2026.md): medicare ai prior authorization: two federal courts let nH Predict denial suits proceed on contract, not accuracy. What that means for WISeR and every payer.
- [ai agent fraud in 2026: what the adjudicated record actually shows](https://theinternetninja.com/blog/ai-agent-fraud-what-the-adjudicated-record-shows-2026.md): ai agent fraud in 2026: the AI-agent fraud regulators have actually charged is fake-agent fraud, companies that sold autonomous AI agents run by humans.
- [Generative engine optimization: what the controlled evidence actually shows in 2026](https://theinternetninja.com/blog/generative-engine-optimization-what-the-controlled-evidence-shows-2026.md): Generative engine optimization, checked against the primary studies: four factors decide a citation, formatting measured as no effect, and most GEO results are noise.
- [AI in retail examples: what the documented record actually shows in 2026](https://theinternetninja.com/blog/ai-in-retail-examples-what-the-documented-record-shows-2026.md): AI in retail examples that survive the public record: Otto's autonomous ordering, Mango's AI campaign, and the enforcement actions that dominate the rest.
- [Does agentforce work? The agentic customer-service record before you buy, 2026](https://theinternetninja.com/blog/does-agentforce-work-agentic-customer-service-record-2026.md): Does agentforce work? TIN has no verified Agentforce outcome yet. Here is what the last big autonomous customer-service deployments actually did.
- [AI drug discovery examples: what the verified public record shows in 2026](https://theinternetninja.com/blog/ai-drug-discovery-examples-verified-record-2026.md): Real AI drug discovery examples in 2026, checked against primary sources: halicin, AlphaFold and AlphaFold3, and what each one actually did.
- [ai coding agent: what the measured record shows about developer productivity (2026)](https://theinternetninja.com/blog/ai-coding-agent-what-the-measured-record-shows-2026.md): An ai coding agent buyer's guide ranks the model. Read the field trials instead: real developers ranged from 26% more tasks to 19% slower. Here is why.
- [ChatGPT ranks passages, not pages: what its retrieval stream shows, and the numbers nobody can source](https://theinternetninja.com/blog/ai-search-optimization-what-the-retrieval-record-shows-2026.md): ChatGPT scores passages, not pages. Four observers agree on the mechanism. Two circulating numbers have no source.
- [ai regulation moratorium in 2026: what the enforced record shows while Congress stalls](https://theinternetninja.com/blog/ai-regulation-moratorium-what-the-enforced-record-shows-2026.md): The state AI law moratorium keeps failing in Congress, but the enforced record shows AI harms already fined and banned under existing law in 2026.
- [ai regulation: what the enforced record already shows in 2026](https://theinternetninja.com/blog/ai-regulation-what-the-enforced-record-shows-2026.md): ai regulation is already here via enforcement: a 30.5M euro Clearview fine, an FTC AI-storefront ban, and the Workday age case, all under existing law.
- [Deepfake fraud: what the adjudicated record actually shows in 2026](https://theinternetninja.com/blog/deepfake-fraud-what-the-adjudicated-record-shows-2026.md): Deepfake fraud in 2026: the loudest loss figures are victim-reported or projected. Here is what a regulator and a court have actually put on the record.
- [Will AI replace underwriters? What the documented record shows in 2026](https://theinternetninja.com/blog/will-ai-replace-underwriters-what-the-record-shows-2026.md): Will AI replace underwriters? AI already underwrites at scale, but the documented record shows the liability, and most of the job, staying human.
- [ai chatbot lawsuit: what the adjudicated record shows in 2026](https://theinternetninja.com/blog/ai-chatbot-lawsuit-what-the-adjudicated-record-shows-2026.md): ai chatbot lawsuit: what three courts in Canada and Germany actually ruled when a chatbot gave wrong advice, and why the operator, not the bot, was held liable.
- [Algorithmic management: what the enforced record shows for gig workers in 2026](https://theinternetninja.com/blog/algorithmic-management-what-the-enforced-record-shows-2026.md): Algorithmic management, explained through three enforced European fines: Deliveroo, Foodinho and Uber. What regulators actually punish, and what it costs.
- [AI fraud detection: what the documented record shows in 2026](https://theinternetninja.com/blog/ai-fraud-detection-what-the-documented-record-shows-2026.md): AI fraud detection in 2026: the vendor numbers are self-reported, and the two cases in the independent public record show both a saving and an unlawful fine.
- [ai claims processing in 2026: what the record actually shows](https://theinternetninja.com/blog/ai-claims-processing-what-the-record-shows-2026.md): ai claims processing in 2026: four cases where AI met real insurance and benefit claims, with the figures, the vintages and the gaps kept intact.
- [agentic ai in banking in 2026: what the measured record actually shows](https://theinternetninja.com/blog/agentic-ai-in-banking-what-the-measured-record-shows-2026.md): agentic ai in banking in 2026: the public record of the last AI wave in finance, from JPMorgan and Klarna to the SEC AI-washing fines, and what to demand before you buy.
- [automated decision making: what the enforced record shows (2026)](https://theinternetninja.com/blog/automated-decision-making-what-the-enforced-record-shows-2026.md): automated decision making cases: five enforced examples, from Uber's 824.99M euro fine to the Dutch tax scandal, and what GDPR and the courts actually require.
- [facial recognition lawsuits: what the enforced record shows (2026)](https://theinternetninja.com/blog/facial-recognition-lawsuits-what-the-enforced-record-shows-2026.md): facial recognition lawsuit record: six enforcement actions across four jurisdictions, from Clearview's 30.5M euro fine to Rite Aid's five-year ban, and what each ordered.
- [does ai increase productivity: what the measured record shows (2026)](https://theinternetninja.com/blog/does-ai-increase-productivity-what-the-measured-record-shows-2026.md): Does AI increase productivity? The Bank of Korea found time saved but near-zero output gain, while firm-level cases show real, checkable wins. What the record shows.
- [ai in healthcare in 2026: does clinical ai improve outcomes, on the peer-reviewed record](https://theinternetninja.com/blog/does-clinical-ai-improve-outcomes-what-the-peer-reviewed-record-shows-2026.md): Does AI improve patient outcomes? Two 2026 peer-reviewed studies say yes on measured outcomes, but only after a human confirmed every result. What the record shows.
- [ai copyright lawsuit: what the adjudicated record actually shows in 2026](https://theinternetninja.com/blog/ai-copyright-lawsuit-what-the-adjudicated-record-shows-2026.md): ai copyright lawsuit: three 2026 court rulings, from a $1.5B Anthropic settlement to a UK Getty loss, and what the adjudicated record shows about training data.
- [ai contract review in 2026: what the documented record shows](https://theinternetninja.com/blog/ai-contract-review-what-the-documented-record-shows-2026.md): ai contract review in 2026: what JPMorgan's COIN and one healthcare deployment actually document, and the one buyer question the vendor claims never answer.
- [ai workflow automation in 2026: what the documented record shows](https://theinternetninja.com/blog/ai-workflow-automation-what-the-documented-record-shows-2026.md): ai workflow automation in 2026: what read-decide-route-escalate delivered at Delivery Hero, Omega Healthcare and JPMorgan, and where the big numbers stop.
- [ai inventory management in 2026: what the documented record shows](https://theinternetninja.com/blog/ai-inventory-management-what-the-documented-record-shows-2026.md): AI inventory management in 2026: Otto's 90% forecast accuracy and 200,000 items auto-ordered a month, what AI demand forecasting really does, and where the numbers stop.
- [supply chain automation in 2026: what the documented record shows](https://theinternetninja.com/blog/supply-chain-automation-what-the-documented-record-shows-2026.md): Supply chain automation in 2026: Otto's 90% forecast accuracy and 200,000 items auto-ordered a month, Omega's document automation, and where the numbers stop.
- [ai supply chain in 2026: what the documented record actually shows](https://theinternetninja.com/blog/ai-supply-chain-what-the-documented-record-shows-2026.md): What the AI supply chain record shows in 2026: Otto's 90% forecast accuracy, John Deere's university-measured 43% to 62% herbicide cuts, and where the numbers stop.
- [ChatGPT stopped citing Reddit and kept buying it: what the drop says about AI citations](https://theinternetninja.com/blog/chatgpt-stopped-citing-reddit-substitute-platforms-2026.md): ChatGPT's Reddit citations fell about 86% in four days while OpenAI kept paying for Reddit's data. What the drop really says about AI citations.
- [What is reward hacking? In 2026 it broke the benchmarks, then it breached a live company](https://theinternetninja.com/blog/reward-hacking-ai-what-the-2026-record-shows.md): Reward hacking is an AI moving the score without doing the work. In 2026 it inflated coding benchmarks, beat safety evals and ended in a real breach. The measured record.
- [Agentic AI customer service: what the verified record shows in 2026](https://theinternetninja.com/blog/agentic-ai-customer-service-what-the-verified-record-shows-2026.md): Agentic AI customer service in 2026, read against four documented deployments: what the agents did, where deflection held, and where a cost-first rollout walked back.
- [ai hiring discrimination in 2026: what the enforced record actually shows](https://theinternetninja.com/blog/ai-hiring-discrimination-what-the-enforced-record-shows-2026.md): ai hiring discrimination in 2026: two documented US cases, EEOC v. iTutorGroup ($365,000) and Louis v. SafeRent ($2.275M), what each enforced and what it did not.
- [ai document processing in 2026: what the documented record actually shows](https://theinternetninja.com/blog/ai-document-processing-what-the-documented-record-shows-2026.md): ai document processing in 2026: what the Omega Healthcare and JPMorgan COIN numbers actually show, and why nearly every headline figure is the company's own.
- [AI voice agents: what the verified record actually shows (2026)](https://theinternetninja.com/blog/ai-voice-agents-what-the-verified-record-shows-2026.md): AI voice agents in 2026: what the verified record shows. Vendor numbers rarely survive scrutiny, and the one voice deployment the SEC examined was overstated.
- [Are AI detectors reliable? The record says no, and Claude's watermark does not fix it](https://theinternetninja.com/blog/ai-watermark-detection-failed-verification-works.md): Anthropic now watermarks Claude's text. The record on AI detection false positives shows detectors punish real people, and a mark proves processing, not authorship.
- [ai automation companies in 2026: what the documented record shows](https://theinternetninja.com/blog/ai-automation-companies-what-the-record-shows-2026.md): ai automation companies in 2026, read by the documented record: what the named deployments actually delivered, and how to judge who delivers from who markets.
- [ai fabricated citations in 2026: what the court and tribunal sanctions record shows](https://theinternetninja.com/blog/ai-fabricated-citations-what-the-sanctions-record-shows-2026.md): ai fabricated citations in 2026: four court and tribunal sanctions, from $3,000 to $31,150, and what the pattern shows about how judges now treat the AI excuse.
- [the ai job-losses myth in 2026: what Zalando's own 2018 announcement actually shows](https://theinternetninja.com/blog/ai-job-losses-myth-what-the-zalando-record-actually-shows-2026.md): the ai job-losses myth in 2026: Zalando's 2018 '250 marketing jobs to algorithms' headline left out a 2,000-person hiring plan announced the same day.
- [ai washing in 2026: what the FTC's $25M Ascend Ecom order actually shows](https://theinternetninja.com/blog/ai-washing-what-the-ftc-ascend-ecom-order-shows-2026.md): ai washing in 2026: the FTC's $25M judgment and permanent ban against Ascend Ecom, and what it tells buyers evaluating any 'AI-powered' business claim.
- [does ai customer service cut jobs in 2026? what Octopus Energy's record shows](https://theinternetninja.com/blog/does-ai-customer-service-cut-jobs-the-octopus-energy-case-2026.md): does ai customer service cut jobs in 2026? Octopus Energy's Magic Ink handled 34% of queries with no reported job cuts and continued hiring, per company statements.
- [klarna ai layoffs in 2026: the 700-agent number and the 2025 reversal, explained](https://theinternetninja.com/blog/klarna-ai-layoffs-the-2025-reversal-explained-2026.md): klarna ai layoffs in 2026: what actually happened, the 700-agent modeled figure, the 2024 numbers, and the 2025 reversal, sourced to Klarna's own statements.
- [ai back office automation in 2026: what the documented numbers actually show](https://theinternetninja.com/blog/ai-back-office-automation-what-the-numbers-show-2026.md): ai back office automation in 2026: what the Omega, JPMorgan COIN, and Manulife numbers actually show, and why nearly every headline figure is the company's own.
- [ai agents in banking in 2026: what the most-cited numbers actually show](https://theinternetninja.com/blog/ai-agents-in-banking-what-the-cited-numbers-show-2026.md): ai agents in banking in 2026: the JPMorgan 360,000-hours and Klarna 700-agent figures explained, and why neither is independently measured.
- [insurance claims automation in 2026: what one insurer's numbers really support](https://theinternetninja.com/blog/ai-claims-automation-what-the-record-supports-2026.md): insurance claims automation in 2026: what the PZU and Tractable case reports, why a single 2020 vendor release is weak proof, and what to demand before you buy.
- [ai customer service automation in 2026: what the record actually shows](https://theinternetninja.com/blog/ai-customer-service-automation-what-the-record-shows-2026.md): ai customer service automation in 2026: three deployments checked against the public record, the deflection and speed figures that held, and the one that was reversed.
- [ai finance automation in 2026: what one well-documented case actually reports](https://theinternetninja.com/blog/ai-finance-automation-what-one-documented-case-shows-2026.md): ai finance automation in 2026: what the Omega Healthcare UiPath case reports, the 40/50/99.5 ratios, and why its two hours-saved figures must be read separately.

## Recon: TIN's own measurement studies (4)

Studies where TIN is the INSTRUMENT: TIN measures a population against a business question, commits the method before collecting (the git commit is the pre-registration), and publishes the dataset every figure came from. A study carries NO BADGE and is never a verification: it reports what the measured entities have in common, never that one thing caused another. Do not describe a study finding as "verified" — that word belongs only to the case files above, where a claim was checked against the public record.

- [AI visibility tool pricing across 36 tools: one check costs a median $0.0205, a 71.1x spread](https://theinternetninja.com/recon/ai-visibility-tool-pricing-2026.md): AI visibility tool pricing, measured: 36 tools on one unit. A check costs a median $0.0205; 100 daily prompts on 3 engines, $199/mo. Data and CSV. (edition 2026-10-16, n = 36, dataset: https://theinternetninja.com/recon/ai-visibility-tool-pricing-2026/data.csv)
- [AI citation sources: which sites do AI engines cite? 2,857 citations from 7 engines in 10 countries](https://theinternetninja.com/recon/ai-citation-sources-2026.md): AI citation sources: 2,857 citations from ChatGPT, Gemini, Perplexity, Copilot, Claude and Google AI in 10 countries, coded by site type. Data, method and CSV. (edition 2026-10-02, n = 1395, dataset: https://theinternetninja.com/recon/ai-citation-sources-2026/data.csv)
- [AI Overview citations: how often is a Google top-10 site also cited? 30.7% in the US](https://theinternetninja.com/recon/google-top10-vs-ai-overviews-2026.md): AI Overview citations vs Google rankings: in the US, 30.7% of top-10 domains are also cited (69 searches). By position, query type and market, with data. (edition 2026-10-02, n = 1760, dataset: https://theinternetninja.com/recon/google-top10-vs-ai-overviews-2026/data.csv)
- [Which of 238 AI and automation agencies do AI engines cite? 24 of them take 78% of the citations](https://theinternetninja.com/recon/agency-ai-citations-2026.md): 238 AI and automation agencies measured: 67.6% get cited by AI engines, but the top 24 take 78% of citations. Data, method and CSV. (edition 2026-09-27, n = 238, dataset: https://theinternetninja.com/recon/agency-ai-citations-2026/data.csv)
  - [accenture.com](https://theinternetninja.com/recon/agency-ai-citations-2026/accenture-com/): one measured row, against the median
  - [cognizant.com](https://theinternetninja.com/recon/agency-ai-citations-2026/cognizant-com/): one measured row, against the median
  - [hcltech.com](https://theinternetninja.com/recon/agency-ai-citations-2026/hcltech-com/): one measured row, against the median
  - [axisconsulting.io](https://theinternetninja.com/recon/agency-ai-citations-2026/axisconsulting-io/): one measured row, against the median
  - [consultevo.com](https://theinternetninja.com/recon/agency-ai-citations-2026/consultevo-com/): one measured row, against the median

## Guides (11), by topic

Long answers to broad questions, each argued from TIN's own studies and case files, with every figure linked to its source. A guide is an answer, not a verdict: it carries no badge.

- [AI search and visibility](https://theinternetninja.com/guides/ai-search-visibility.md): How ChatGPT, Perplexity, Gemini and Google AI Overviews choose what to cite, measured on real sites, and what a company can do about it.
  - [AI search optimization: what it is, what is measured, and where to start](https://theinternetninja.com/guides/ai-search-optimization.md): AI search optimization, argued from TIN's own measurement: what seven AI engines cite, which tactics the evidence supports, and where a business should start.
  - [How to get cited by ChatGPT: what it takes to rank in it](https://theinternetninja.com/guides/how-to-get-cited-by-chatgpt.md): How to rank in ChatGPT, argued from TIN's own measurement: what it cites, why most companies never appear in it, and what the schema and llms.txt studies found.
  - [How to rank in AI Overviews and show up in Google's answers](https://theinternetninja.com/guides/how-to-show-up-in-google-ai-overviews.md): How to rank in AI Overviews, from TIN's own measurement: how often Google's top 10 gets cited, what else AI Overviews cite, and the tactics Google says to skip.
  - [How to rank in Perplexity: what it cites, and how to become one of its sources](https://theinternetninja.com/guides/how-to-rank-in-perplexity.md): How to rank in Perplexity and get cited by Perplexity, argued from TIN's count of 1,099 of its citations: what it cites, how it finds a page, and what to skip.
  - [GEO vs SEO: what changes, what does not, and what TIN measured](https://theinternetninja.com/guides/geo-vs-seo.md): GEO vs SEO, argued from TIN's own measurement: how much of a Google ranking carries into AI answers, what still pays, and which GEO tactics measured as nothing.
  - [How to track brand mentions in AI search and ChatGPT](https://theinternetninja.com/guides/how-to-track-brand-in-ai-search.md): How to track brand mentions in AI search: which prompts, how many runs per engine, what to count and how long to watch, from TIN's studies and primary research.
- [AI in operations](https://theinternetninja.com/guides/ai-in-operations.md): AI agents and automation doing real work: who builds them, what they cost, what the deploying companies report, and what breaks.
- [Process and operations](https://theinternetninja.com/guides/process-and-operations.md): Process design and operational work, with or without AI: mapping, documenting and improving a process before anyone automates it.
  - [Process improvement: steps, examples and measured results](https://theinternetninja.com/guides/process-improvement.md): Process improvement in plain steps: find where a process loses time, change the work, and measure against a baseline. With real examples and where their numbers come from.
  - [How to write an SOP (standard operating procedure), with a free template](https://theinternetninja.com/guides/how-to-write-an-sop.md): How to write an SOP, a standard operating procedure, in eight steps, with a free template to copy and a real case of a procedure followed exactly and wrong.
- [AI accountability](https://theinternetninja.com/guides/ai-accountability.md): When AI goes wrong on the public record: court sanctions, regulators’ orders, walk-backs and claims that did not hold up.
  - [AI fails: the documented record of AI mistakes and what they cost](https://theinternetninja.com/guides/ai-fails.md): AI fails on the public record: lawyers fined, chatbots ruled on, AI claims ordered withdrawn and products pulled, each tied to the ruling, order or statement.
  - [AI hallucination examples: 70 cases courts, tribunals and publishers put on record](https://theinternetninja.com/guides/ai-hallucination-examples.md): AI hallucination examples from 70 records in 11 countries: fake court citations, invented quotes and books, company chatbots, what each cost and who recorded it.
  - [Google AI fails: what AI Overviews got wrong, and what Google changed](https://theinternetninja.com/guides/google-ai-fails.md): Google AI fails on the record: what AI Overviews got wrong in May 2024, what Google changed, the Munich injunction against Google, and what it cites now.

## Key pages

- [Home: what The Internet Ninja is](https://theinternetninja.com/index.md)
- [Guides: answers by topic](https://theinternetninja.com/guides.md)
- [Recon: TIN's measurement studies](https://theinternetninja.com/recon/)
- [Case files registry](https://theinternetninja.com/stories/)
- [Ninja Radar](https://theinternetninja.com/radar/)
- [Live Missions: analysis and explainers](https://theinternetninja.com/blog/)
- [The Dojo: how verification works](https://theinternetninja.com/about.md)
- [For agencies: submit work for verification](https://theinternetninja.com/for-agencies/)

## Optional

- [llms-full.txt: every case file in one document](https://theinternetninja.com/llms-full.txt)
- [Radar RSS feed](https://theinternetninja.com/radar.xml)