# A benefits algorithm brought down a government: the Dutch childcare-benefits scandal and its EUR 2.75m discrimination fine

> The Dutch Tax Administration used applicants' nationality as an indicator in a system that automatically flagged childcare-benefit claims as risky. On 7 December 2021 the Dutch Data Protection Authority fined the state EUR 2.75 million, finding it had processed nationality data 'in an unlawful, discriminatory and therefore improper manner' in breach of the GDPR. The wider scandal (the 'toeslagenaffaire'), in which a parliamentary inquiry found families wrongly accused of fraud in an 'unprecedented injustice', brought down the third Rutte cabinet, which resigned on 15 January 2021. This is a verified negative about a live government algorithm, documented by the regulator and the record rather than by the department that ran it.

- Verification status: pending
- Case type: deployment
- Provider: Dutch Tax and Customs Administration (Belastingdienst / Dutch Tax Administration), Ministry of Finance — operator of the childcare-benefit risk-classification system that used nationality as a risk indicator; no commercial vendor exists in this honest-negative
- Client: Dutch Tax Administration / Ministry of Finance (the Dutch state) — the deploying government body and subject; adjudicated by the Autoriteit Persoonsgegevens, and the wider affair brought down the third Rutte cabinet, government (named)
- Sector: government / NL / cross
- Canonical URL: https://theinternetninja.com/stories/netherlands-toeslagenaffaire-childcare-benefits-risk-algorithm-dpa-2-75m-fine-2021/
- Source: The Internet Ninja (theinternetninja.com), independent verified-proof platform

## Outcomes

| Metric | Before | After |
| --- | --- | --- |
| The fine |  |  |
| The mechanic |  |  |
| The scale |  |  |
| The fall |  |  |

## Verification method

Every load-bearing figure is quoted verbatim from sources fetched live this session (2026-09-02), each with a web.archive.org snapshot. The EUR 2.75m fine, the unlawful/discriminatory finding, the nationality-as-risk-indicator mechanic and the scale figures are from the Dutch DPA's own announcement (Tier 1 primary), the fine and finding also reported independently by NL Times (Tier 2). The algorithmic characterisation is corroborated by Amnesty International's 'Xenophobic machines' report (Tier 2). The 15 January 2021 cabinet resignation is confirmed by the Dutch government's own official resignation statement (Rijksoverheid, Tier 1 primary), which names the childcare-benefits affair and is the source of the 'unprecedented injustice' phrasing, and is further corroborated by two independent international outlets (CNBC and Al Jazeera, Tier 2). No confirmation was sought from any party; only already-public records are used. Awaits the checker's audit.

## Full case file

**Verification status: CHECKING — every quoted figure is taken verbatim from the Dutch Data Protection Authority's own announcement (Tier 1 primary, with NL Times corroborating the fine independently) and from the Dutch government's own resignation statement (Tier 1 primary) plus contemporaneous independent reporting of the January 2021 cabinet resignation (CNBC, Al Jazeera, Tier 2), corroborated by Amnesty International's investigation. This is a verified negative about a live government algorithm; no green badge is claimed, and the story awaits the checker's audit.**

## The problem
The Dutch state did not buy this system from a vendor; it ran nationality through the machinery of its own benefits administration. The Dutch Tax Administration processed data on the (dual) nationality of childcare-benefit (kinderopvangtoeslag) applicants across several of its processes, and one of them was a risk filter: "it used applicants' nationality (Dutch/not Dutch) as an indicator in a system that automatically designated certain applications as risky" ([source](https://www.autoriteitpersoonsgegevens.nl/en/current/tax-administration-fined-for-discriminatory-and-unlawful-data-processing)). Amnesty International, investigating the same system independently, described it in the language of automated decision-making: "The Dutch tax authorities adopted an algorithmic decision-making system to create risk profiles of individuals applying for childcare benefits in order to detect inaccurate and potentially fraudulent applications at an early stage. Nationality was one of the risk factors used by the tax authorities" ([source](https://www.amnesty.org/en/documents/eur35/4686/2021/en/)). Amnesty found that "the use of individuals' nationality resulted in discrimination as well as racial profiling" ([source](https://www.amnesty.org/en/documents/eur35/4686/2021/en/)). This is a verified negative about a government AI system, documented by the regulator and an independent watchdog rather than by the department that ran it.

## What the algorithm did
Nationality was not incidental to the design; it was a factor the system was never entitled to use. The regulator was explicit that the data was unnecessary for the purpose: after describing the automatic risk-designation, the DPA held that "the data was not necessary for this purpose either" and that "it is unlawful, and therefore prohibited, to use nationality data to assess applications, combat fraud and determine risk" ([source](https://www.autoriteitpersoonsgegevens.nl/en/current/tax-administration-fined-for-discriminatory-and-unlawful-data-processing)). The DPA drew the line from the algorithm to the harm directly: "unlawful processing by means of an algorithm led to a violation of the right to equality and non-discrimination" ([source](https://www.autoriteitpersoonsgegevens.nl/en/current/tax-administration-fined-for-discriminatory-and-unlawful-data-processing)). The scale of the retained data was large and stale: the Tax Administration "should have deleted the data on dual nationality of Dutch nationals back in January 2014," yet "in May 2018, some <span class="kpi">1.4 million</span> people were still registered as dual nationals in its systems" ([source](https://www.autoriteitpersoonsgegevens.nl/en/current/tax-administration-fined-for-discriminatory-and-unlawful-data-processing)).

## The outcome
The regulator imposed one of its heaviest fines on the state itself. "The Dutch Data Protection Authority (DPA) has imposed a <span class="kpi">€2.75 million</span> fine on the Dutch Tax Administration," because "for many years the Tax Administration processed data on the (dual) nationality of childcare benefit applicants in an unlawful, discriminatory and therefore improper manner," a "serious violation[] of the General Data Protection Regulation (GDPR)" ([source](https://www.autoriteitpersoonsgegevens.nl/en/current/tax-administration-fined-for-discriminatory-and-unlawful-data-processing)). DPA chair Aleid Wolfsen tied the case to the wider affair: "That went horribly wrong at the Benefits Office, with all the associated consequences. Obviously this fine cannot undo any of the harm done. But it is an important step within a broader recovery process" ([source](https://www.autoriteitpersoonsgegevens.nl/en/current/tax-administration-fined-for-discriminatory-and-unlawful-data-processing)). The department had by then stopped: "The Tax Administration has not used applicants' nationality in determining risk since <span class="kpi">October 2018</span>. And it has not used the data to combat organised fraud since <span class="kpi">February 2019</span>" ([source](https://www.autoriteitpersoonsgegevens.nl/en/current/tax-administration-fined-for-discriminatory-and-unlawful-data-processing)). The fine and the finding were reported independently the next day: NL Times wrote that "the Tax Authority will be fined <span class="kpi">2.75 million euros</span> for the childcare allowance scandal," and that "according to the regulator, the Tax Authority 'for years processed the (dual) nationality of applicants for childcare allowance in an unlawful, discriminatory, and therefore improper manner'" ([source](https://nltimes.nl/2021/12/08/tax-office-fined-eu275-million-privacy-violations-benefits-scandal)).

The political cost was larger than the fine. "The whole Dutch government collectively resigned on Friday after a scandal involving the mismanagement of childcare funds, which drove thousands of families into financial hardship," after an inquiry found "tax officials wrongly accused thousands of working families of fraud and ordered them to repay childcare benefits between 2013 and 2019," an affair Dutch lawmakers called an "unprecedented injustice" ([source](https://www.cnbc.com/2021/01/15/dutch-government-resigns-after-childcare-benefits-scandal-.html)). Prime Minister Mark Rutte framed the failure in constitutional terms: "The rule of law must protect its citizens from an all-powerful government, and here that's gone terribly wrong" ([source](https://www.aljazeera.com/news/2021/1/15/dutch-pm-rutte-and-his-government-quit-over-child-welfare-scandal)). The same parliamentary inquiry "said approximately <span class="kpi">10,000</span> families had been forced to repay tens of thousands of euros of subsidies, in some cases leading to unemployment, bankruptcies and divorces" ([source](https://www.aljazeera.com/news/2021/1/15/dutch-pm-rutte-and-his-government-quit-over-child-welfare-scandal)). The government's own record confirms it: in its official statement, Rutte said the cabinet had failed the norm "in het kinderopvangtoeslagdossier" (in the childcare-benefits file) and that, "als het hele systeem heeft gefaald, kan alleen gezamenlijk verantwoordelijkheid worden gedragen. En dat heeft tot de conclusie geleid dat ik zojuist de Koning het ontslag heb aangeboden van het hele kabinet" (if the whole system has failed, only collective responsibility can be borne, which led to the conclusion that he had just offered the King the resignation of the entire cabinet) ([source](https://www.rijksoverheid.nl/documenten/publicaties/2021/01/15/verklaring-van-minister-president-mark-rutte-over-het-aftreden-van-het-kabinet)). The English phrase "unprecedented injustice" is a translation of the statement's own closing words, that the cabinet acted to do right by the parents "aan wie ongekend onrecht is aangedaan" (to whom unprecedented injustice was done) ([source](https://www.rijksoverheid.nl/documenten/publicaties/2021/01/15/verklaring-van-minister-president-mark-rutte-over-het-aftreden-van-het-kabinet)).

**Weakest load-bearing link.** The EUR 2.75m fine and the unlawful-and-discriminatory finding rest on two independent sources: the regulator itself (the Dutch DPA), a Tier 1 primary that is independent of the Tax Administration it fined, and NL Times reporting the same figure and finding the next day (Tier 2). The 15 January 2021 cabinet resignation is now anchored on a Tier 1 primary, the Dutch government's own official resignation statement, which names the childcare-benefits affair, records the offer of the whole cabinet's resignation to the King, and is the source of the phrase the English press rendered as "unprecedented injustice"; two independent international outlets (CNBC and Al Jazeera) carry the same event and the parliamentary inquiry's characterisation. The full DPA decision PDF and the December 2020 parliamentary report "Ongekend onrecht" would add further primary artifacts, but neither claim now depends on them. Every figure on this page is quoted verbatim from a source fetched this session; none is recalled or rounded.

> **How this was verified.** Method: every load-bearing sentence is quoted verbatim from a source fetched live this session (2026-09-02), each backed by a web.archive.org snapshot. The EUR 2.75m fine, the "unlawful, discriminatory and therefore improper" finding, the nationality-as-risk-indicator mechanic, the 1.4 million dual-national records of May 2018, and the October 2018 / February 2019 cessation dates all come from the Autoriteit Persoonsgegevens (Dutch DPA) announcement of 7 December 2021 (Tier 1 primary). The EUR 2.75m fine and the discriminatory/unlawful finding are additionally corroborated by NL Times, 8 December 2021 (Tier 2). The algorithmic decision-making characterisation and the racial-profiling finding are corroborated by Amnesty International's report "Xenophobic machines" (EUR 35/4686/2021), 25 October 2021 (Tier 2). The 15 January 2021 collective cabinet resignation is confirmed by the Dutch government's own official resignation statement (Rijksoverheid, 15 January 2021, Tier 1 primary), which names the childcare-benefits affair and is the source of the "unprecedented injustice" ("ongekend onrecht") phrasing; the resignation, the "unprecedented injustice" characterisation and the ~10,000-families repayment figure are further corroborated by CNBC and Al Jazeera, both 15 January 2021 (Tier 2). No confirmation was sought from the Tax Administration, the Ministry of Finance or any party; only already-public records are used. Checked 2026-09-02 (round 2).

## Related case files
- [The Hague District Court's SyRI judgment, where a Dutch court declared a government welfare-fraud algorithm unlawful](/stories/hague-district-court-syri-welfare-fraud-algorithm-unlawful-article-8-echr-2020/) — the same country and the same failure mode, decided by a court on Article 8 ECHR rather than by the data regulator.
- [Michigan's MiDAS unemployment-fraud algorithm wrongly accused tens of thousands](/stories/michigan-midas-unemployment-fraud-algorithm-40195-cases-93pct-error-20m-settlement-2024/) — the same failure mode in a US state: a government benefits system that auto-flagged fraud, got it wrong at scale, and cost the state a settlement.
- [A VA watchdog found errors in nearly all 8,100 automated survivors'-benefit decisions it reviewed](/stories/va-oig-automated-survivors-death-benefit-decisions-8000-errors-2-7m-improper-payments-2026/) — another government body letting automation decide benefits eligibility and producing systematic error, documented by an independent oversight body rather than the operator.
- [The US DOJ's Fair Housing settlement over Meta's ad-delivery algorithm](/stories/meta-doj-fair-housing-act-ad-algorithm-vrs-115054-max-penalty-2022/) — a regulator forcing change on an algorithm that discriminated against a protected characteristic, the enforcement-against-algorithmic-discrimination parallel to the DPA's nationality fine.

## Sources
1. Autoriteit Persoonsgegevens (Dutch Data Protection Authority) · *Tax Administration fined for discriminatory and unlawful data processing* · 7 December 2021 · https://www.autoriteitpersoonsgegevens.nl/en/current/tax-administration-fined-for-discriminatory-and-unlawful-data-processing — **Tier 1** (primary; the regulator's own announcement, and the independent source of record for the €2.75m fine, the "unlawful, discriminatory and therefore improper" finding, the nationality-as-risk-indicator mechanic, the "unlawful processing by means of an algorithm" causation, the 1.4 million dual-national records of May 2018, the January 2014 deletion obligation, and the October 2018 / February 2019 cessation dates; archived at web.archive.org/web/20260902040904, excerpts in sources/autoriteitpersoonsgegevens-dpa-fine-2021-12-07.txt).
2. Amnesty International · *Xenophobic machines: Discrimination through unregulated use of algorithms in the Dutch childcare benefits scandal* (Index EUR 35/4686/2021) · 25 October 2021 · https://www.amnesty.org/en/documents/eur35/4686/2021/en/ — **Tier 2** (independent NGO investigative report; corroborates the algorithmic decision-making characterisation, nationality as a risk factor, and the resulting discrimination and racial profiling; archived at web.archive.org/web/20260902041011, excerpts in sources/amnesty-xenophobic-machines-2021-10-25.txt).
3. NL Times · *Tax office fined €2.75 million for privacy violations in benefits scandal* · 8 December 2021 · https://nltimes.nl/2021/12/08/tax-office-fined-eu275-million-privacy-violations-benefits-scandal — **Tier 2** (independent English-language Dutch news outlet; second independent source for the €2.75m fine and the "unlawful, discriminatory, and therefore improper" finding, reported the day after the DPA announcement; archived at web.archive.org/web/20260902044158, excerpts in sources/nltimes-tax-office-fined-2021-12-08.txt).
4. Rijksoverheid (Government of the Netherlands) · *Verklaring van minister-president Mark Rutte over het aftreden van het kabinet* (Statement by PM Mark Rutte on the resignation of the cabinet) · 15 January 2021 · https://www.rijksoverheid.nl/documenten/publicaties/2021/01/15/verklaring-van-minister-president-mark-rutte-over-het-aftreden-van-het-kabinet — **Tier 1** (primary; the Dutch government's own official statement of the collective resignation, naming the kinderopvangtoeslag affair, recording the offer of the whole cabinet's resignation to the King, and the origin of the "ongekend onrecht" / "unprecedented injustice" phrasing; archived at web.archive.org/web/20260310222618, excerpts in sources/rijksoverheid-rutte-resignation-statement-2021-01-15.txt).
5. CNBC · *Dutch government resigns after childcare benefits scandal* · 15 January 2021 · https://www.cnbc.com/2021/01/15/dutch-government-resigns-after-childcare-benefits-scandal-.html — **Tier 2** (independent business press; source for the collective cabinet resignation of 15 January 2021, the December parliamentary investigation finding tax officials wrongly accused thousands of families between 2013 and 2019, and the "unprecedented injustice" characterisation; archived at web.archive.org/web/20251113104800, excerpts in sources/cnbc-dutch-government-resigns-2021-01-15.txt).
6. Al Jazeera · *Dutch PM Rutte and his government quit over child welfare scandal* · 15 January 2021 · https://www.aljazeera.com/news/2021/1/15/dutch-pm-rutte-and-his-government-quit-over-child-welfare-scandal — **Tier 2** (independent international press; corroborates the resignation, carries Rutte's "the rule of law must protect its citizens from an all-powerful government" statement, and the parliamentary inquiry's ~10,000-families / "unprecedented injustice" finding; archived at web.archive.org/web/20260902041026, excerpts in sources/aljazeera-rutte-resignation-2021-01-15.txt).