Skip to content
Submit incident
AII-1000-70603
ConfirmedA court, a regulator or the organization itself has stated the core facts, or the record cites a court or regulator source.

NRW police data analysis law permits insufficient AI safeguards

NRW's revised police law permits the Palantir system to operate with AI and allows police to train models on their own data, including personal data when anonymization would be burdensome. The state data protection commissioner warned that insufficient safeguards govern these practices and that unreliable police databases could cause serious harms when used to train AI systems predicting crime.

17 April 2026 · Polizei Nordrhein-Westfalen · Privacy and surveillance
17 April 2026
AII-1000-706032 sources ↓
Quick facts
Category
Privacy and surveillance
Country
Germany
Source
RP Online
Date
17 April 2026
Drafted by
Raidu Incident Detection

What happened

Under a revised NRW police law passed in 2025, the Palantir data analysis system can now operate using AI, despite previous decisions against this on safety grounds. The law permits police to train AI using their own data, including personal data when anonymization would require disproportionate effort.

Reported impact

Police databases contain unreliable and rarely updated data. A complainant was denied US entry because border control cited a police suspicion that was later disproven by acquittal. When police retain undisproven suspicions, using such data to train AI systems could have severe consequences.

Index analysis

The NRW data protection commissioner warned that insufficient safeguards govern the use and training of AI in both the revised Police Law and the new Constitutional Protection Law. The laws do not adequately address the different impacts of diverse AI applications or the problems arising from defective data bases.

Governance controls

Laws were passed in expedited procedures without sufficient care to protect fundamental rights. The laws permit using unreliable police data for AI training and allow personal data use when anonymization would impose disproportionate effort.

Classification

Category
Privacy and surveillance
Industry
Government, security and defence
Country
Germany
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Human or fundamental rights, Public interest
Harmed
General public
Failure modes
Biased model
AI systems
Palantir
Regulations
GDPR (cited)

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

RP Online
Datenschutz in Nordrhein-Westfalen: NRW-Landesdatenschützerin Gayk meldet Höchststand an Beschwerden
17 April 2026
heise online
NRW-Datenschutzbeauftragte warnt vor Datennutzung um jeden Preis
19 April 2026