# Connecticut's high court, for the first time, sanctions a lawyer for ChatGPT-hallucinated citations

> In a per curiam order released July 31, 2026, the Connecticut Supreme Court, confronting a lawyer's use of generative AI 'for the first time', sanctioned Attorney Ian G. Gottlieb and GLG Law, LLC for filing ~seven ChatGPT-hallucinated citations: six hours of CLE (three on generative AI) and a $1,000 donation each to the CT Bar Institute.

- Verification status: verified
- Case type: deployment
- Provider: ChatGPT (OpenAI)
- Client: Connecticut Supreme Court — Kosel Equity, LLC v. MacGregor (SC 21184) / TOV Realty, LLC v. Suarez (SC 21183), Courts / legal (AI-governance honest-negative) (named)
- Sector: legal / US / cross
- Verified on: 2026-08-26
- Canonical URL: https://theinternetninja.com/stories/ct-supreme-court-tov-realty-kosel-gottlieb-glg-law-chatgpt-hallucinated-citations-1000-donation-cle-first-impression-2026/
- Source: The Internet Ninja (theinternetninja.com), independent verified-proof platform

## Outcomes

| Metric | Before | After |
| --- | --- | --- |
| CLE ordered (beyond the annual 12-hour minimum), three hours of which must concern generative AI: 6 hours |  |  |
| Charitable donation to the CT Bar Institute, Inc. — Attorney Gottlieb and the law firm each: $1,000 each |  |  |
| Erroneous, unverified citations Gottlieb admitted filing 'as a result of the use of generative AI': approximately seven |  |  |

## Verification method

Court-origin: every figure is quoted verbatim from the Connecticut Supreme Court's own signed per curiam reasons (jud.ct.gov ORD355.3.pdf), corroborated firsthand by two independent secondaries — Eugene Volokh / Reason (Aug 4, 2026) and Law360 (July 31, 2026). Two independent secondaries → court-origin two_independent.

## FAQ

**What did the Connecticut Supreme Court do about the ChatGPT-hallucinated citations?**

In a per curiam order officially released July 31, 2026, the Connecticut Supreme Court sanctioned Attorney Ian G. Gottlieb and GLG Law, LLC for filing approximately seven erroneous citations produced when he used ChatGPT to edit otherwise-accurate drafts. The court said it was confronting a lawyer's use of generative AI for the first time.

**What was the sanction in the Gottlieb GLG Law case?**

Gottlieb was ordered to complete six hours of continuing legal education beyond the annual 12-hour minimum, three hours of which must concern generative AI, and he and the firm each had to donate $1,000 to the CT Bar Institute, Inc.

**Was Attorney Gottlieb referred for further discipline?**

No. The order required a compliance report with a courtesy copy to the Statewide Grievance Committee, but the court emphasized that this courtesy copy is not a referral for further disciplinary action.

## Full case file

## What happened

Attorney Ian G. Gottlieb and his firm, GLG Law, LLC, filed applications for certification to bring two public-interest appeals in Connecticut under General Statutes section 52-265a, Kosel Equity, LLC v. MacGregor (SC 21184) and TOV Realty, LLC v. Suarez (SC 21183) ([source](https://www.jud.ct.gov/external/supapp/Cases/AROcr/CR355/ORD355.3.pdf)). Gottlieb "initially used LexisNexis to conduct the legal research necessary to draft his public interest appeal applications and the briefs that followed" ([source](https://reason.com/volokh/2026/08/04/chatgpt-editing-adding-hallucinations-to-otherwise-accurate-material/)). He then "copied and pasted those drafts into ChatGPT, a generative AI program, to make his brief writing process more efficient and to improve the quality of the final written product" ([source](https://www.jud.ct.gov/external/supapp/Cases/AROcr/CR355/ORD355.3.pdf)). Unbeknownst to him, "the drafts produced by ChatGPT added new case citations or altered existing case citations," and those AI-generated changes did not prompt a second round of verification ([source](https://www.jud.ct.gov/external/supapp/Cases/AROcr/CR355/ORD355.3.pdf)).

In a per curiam order officially released July 31, 2026, the Connecticut Supreme Court stated it was "required to confront for the first time a lawyer's use of generative artificial intelligence (generative AI) that has resulted in the submission of written materials to this court containing citations that were 'hallucinated' — in other words, fabricated or inaccurately generated" ([source](https://www.jud.ct.gov/external/supapp/Cases/AROcr/CR355/ORD355.3.pdf)). At a show-cause hearing held July 7, 2026, Gottlieb "candidly admitted that his conduct, in filing documents that contained approximately seven erroneous and unverified citations as a result of the use of generative AI violated, at a minimum, rule 1.1 of the Rules of Professional Conduct" ([source](https://www.jud.ct.gov/external/supapp/Cases/AROcr/CR355/ORD355.3.pdf)). The count of hallucinated citations must be read exactly as the Court framed it, "approximately seven," not a hard number ([source](https://www.jud.ct.gov/external/supapp/Cases/AROcr/CR355/ORD355.3.pdf)).

## The sanction

The Court ordered that, "in addition to the twelve (12) hours of minimum continuing legal education required annually by Practice Book § 2-27A, Attorney Gottlieb shall complete six (6) hours of continuing legal education in the areas of ethics and law office management, three (3) hours of which must concern the use of generative AI" ([source](https://www.jud.ct.gov/external/supapp/Cases/AROcr/CR355/ORD355.3.pdf)), terms an independent Law360 newsroom reported firsthand as the court ordering Gottlieb to "complete six hours of continuing legal education … three hours of which must concern the use of generative artificial intelligence" ([source](https://www.law360.com/legalethics/articles/2508541)). It further ordered that "Attorney Gottlieb and the law firm each shall donate $1000 to the CT Bar Institute, Inc., to support its efforts to provide the legal profession with instruction in the areas of legal ethics and law office management, and regarding the role of generative AI in those areas," a $2,000 charitable total ([source](https://www.jud.ct.gov/external/supapp/Cases/AROcr/CR355/ORD355.3.pdf)). In mitigation the Court noted Gottlieb filed errata sheets, "showed sincere contrition, and he has no disciplinary history," adding "Nevertheless, the submission of erroneous citations in violation of the Rules of Professional Conduct is not something we take lightly" ([source](https://www.jud.ct.gov/external/supapp/Cases/AROcr/CR355/ORD355.3.pdf)). The order requires a compliance report with a courtesy copy to the Statewide Grievance Committee, but the Court emphasized "this courtesy copy is not a referral for further disciplinary action" ([source](https://www.jud.ct.gov/external/supapp/Cases/AROcr/CR355/ORD355.3.pdf)).

The unusual mechanism is worth preserving: the citations were correct until the attorney used ChatGPT to *edit* otherwise-accurate drafts, and the AI injected hallucinations during that editing pass, distinct from the more familiar pattern of asking a chatbot to find cases from scratch. Eugene Volokh, writing at The Volokh Conspiracy (Reason), independently named the case, the court, Attorney Gottlieb, the ChatGPT-editing mechanism and the full sanction firsthand under his own analytic headline ([source](https://reason.com/volokh/2026/08/04/chatgpt-editing-adding-hallucinations-to-otherwise-accurate-material/)).

**Weakest load-bearing source.** Beyond the Court's own per curiam reasons (Tier 1), two genuinely independent secondaries carry the sanction terms firsthand: Eugene Volokh / Reason (Tier 2), which largely reproduces the opinion text rather than re-reporting it, and Law360 (Tier 2), original Portfolio Media/LexisNexis newsroom reporting that paraphrases the July-31 per curiam order in the reporter's own framing and states both the six-hours-CLE and the $1,000-each donation firsthand. Law360's article sits behind a subscription wall for ordinary browsers, so the weaker access point is that its byte-tie rests on a desktop-user-agent fetch plus a Wayback snapshot (capture 20260820232958) rather than a freely readable page. With Volokh reprinting and Law360 re-reporting, this story is court-origin *two independent*.

## How this was verified

Court-origin verification, settled 2026-08-20. Every figure is quoted verbatim from the Connecticut Supreme Court's own signed per curiam reasons (jud.ct.gov `ORD355.3.pdf`), corroborated firsthand by two genuinely independent secondaries, Eugene Volokh / Reason (Aug 4, 2026) and Law360 (July 31, 2026), so court-origin *two_independent* on the sanction terms. The official primary PDF was archived to the Wayback Machine (capture 20260819225650); its CDX digest `S4IWSAF6TBM7FSVP7ELHPCLZAIEB3QOL` matches the SHA-1 of the live primary byte-for-byte, tying the archived copy to the court's original. The Law360 report was captured live (desktop-UA fetch) and to the Wayback Machine (capture 20260820232958). No subject was asked to confirm any figure; the numbers are court-adjudicated.

## Sources

1. Connecticut Supreme Court (Connecticut Judicial Branch) · *Kosel Equity, LLC v. MacGregor (SC 21184) / TOV Realty, LLC v. Suarez (SC 21183) — per curiam order* · July 31, 2026 · https://www.jud.ct.gov/external/supapp/Cases/AROcr/CR355/ORD355.3.pdf — **Tier 1** (primary; the court's own signed reasons).
2. Eugene Volokh, The Volokh Conspiracy (Reason) · *ChatGPT Editing Adding Hallucinations to Otherwise Accurate Material?* · August 4, 2026 · https://reason.com/volokh/2026/08/04/chatgpt-editing-adding-hallucinations-to-otherwise-accurate-material/ — **Tier 2** (independent legal press naming the parties and the sanction firsthand; largely reprints the opinion).
3. Law360 (Portfolio Media / LexisNexis) · *Conn. Justices Order Education For Atty Who Missed AI Flubs* · July 31, 2026 · https://www.law360.com/legalethics/articles/2508541 — **Tier 2** (independent newsroom; original reporting paraphrasing the per curiam order, carrying both the six-hours-CLE and $1,000-each terms firsthand; subscription wall, byte-tied via desktop-UA fetch + Wayback capture 20260820232958).

## Related case files

- [another Connecticut federal court sanctioning AI-fabricated citations, with a grievance referral this order pointedly withholds](/stories/barteca-v-tacobarn-d-conn-oliver-miller-3500-ai-fabricated-citations-grievance-referral-2026/)
- [a federal appeals court's own "first impression" AI-citation sanction, decided the same summer](/stories/mccarthy-v-dea-3d-cir-24-2704-ai-fabricated-citations-public-reprimand-first-impression-2026/)
- [another state supreme court striking an AI-fabricated brief and referring the lawyer for discipline](/stories/prososki-v-regan-nebraska-supreme-court-lake-ai-fictitious-citations-brief-struck-appeal-dismissed-discipline-referral-2026/)