AI Risk Wire

Cases › biased or discriminatory › LR-2026-0783

EU bank AML algorithms flag minority names in 61.8% of alerts

Preliminary Why: 1 report: Tech Times · subject to change · 1 independent report

Status: no action reported

Severity moderate
Reported 11 Mar 2026
Industry banking

No source states: loss or records, developer and model, agent tool, could do, deployer.

The subject is bank anti-money-laundering screening in the Netherlands, where suspicious-activity reports are produced for customers. A Netherlands Court of Audit review published March 11, 2026 examined 217 such reports and found 61.8% involved a person with a foreign-sounding surname, an overrepresentation the court said could not be explained by any documented risk factor. The same review found mosques questioned more often than Catholic and Protestant institutions. In mid-2024 the Netherlands Institute for Human Rights ruled against ING after customers' transactions were repeatedly blocked or delayed. The sources do not state that an algorithm produced the audited reports, nor a root cause. In May 2026 the Dutch State Commission against Discrimination and Racism called for a halt to data-driven profiling for fraud and crime detection. The EU anti-money-laundering authority is finalising screening guidelines expected in Q4 2026.

What happened

1 · What it was doingDutch banks filed large volumes of unusual-transaction reports with the national financial intelligence unit.
2 · What went wrongReports examined showed a high share of people with foreign-sounding surnames that the court could not explain by documented risk factors.
3 · What it reached or didTwo ING customers had transactions repeatedly blocked or delayed; the human rights institute ruled against the bank.
4 · How it was caughtA Netherlands Court of Audit review published the findings in March 2026.

What was done

No source states an action or fix.

Risks

Bias and discrimination
Primary12 incidents

Read from the sources; not yet reviewed by a person.

Controls

Not yet read for controls. Controls that mitigate the primary risk:

Application

Bias testing · Risk assessment per deployment · Output filtering and redaction

Evidence

No source states: when it happened, how many records were exposed, the direct loss, a count of the other parties affected, any court case, the cause, what set it off, what was done.

Show the full account: 8 items from the primary sources

What happened, in order

  1. 2020 Dutch banks reported nearly 250,000 unusual transactions to FIU-NL.
  2. 2024 Reported unusual transactions had more than doubled to more than 530,000.
  3. mid-2024 Netherlands Institute for Human Rights ruled against ING over repeated blocked or delayed transactions.
  4. Q4 2026 AMLA expected to issue final guidelines.
  5. May 2026 Dutch State Commission against Discrimination and Racism called for a halt to data-driven profiling.
  6. June 2026 Amnesty International published Automating Suspicion on risk profiling systems.
  7. September 2, 2026 Privacy First submitted a consultation response to AMLA.
  8. July 10, 2027 The Anti-Money Laundering Regulation takes effect.
Show every field, with its source and grade

What happened

Earliest source we hold
2026-03-11
C one secondary source states it
What went wrong
biased or discriminatory
C one secondary source states it
primary risk
RK16
D inferred by us from the described behaviour, reason shown

Impact

Severity
moderate
D inferred by us from the described behaviour, reason shown
Kind of harm
discrimination
C one secondary source states it

Context

Industry
banking
C one secondary source states it
Country
Netherlands
C one secondary source states it

Notes on gaps

No date of occurrence is stated. The 217-report sample is the only stated count; affected parties were not counted from the 61.8% figure. No remediation by a party is stated; AMLA guidelines and Privacy First requests are proposals, not changes made, so remediations are empty. The DNB fine on ABN AMRO and the Amnesty and State Commission items concern other matters and are included only as dated events in the timeline. The Dutch Banking Association acknowledgement and the Bank Policy Institute estimates were not included as values.

LR-2026-0783 · version 1 · published 2026-11-01 after automatic checks; not yet checked field by field by a person · JSON