Skip to content
Submit incident
AII-1001-56607
ConfirmedA court, a regulator or the organization itself has stated the core facts, or the record cites a court or regulator source.Supporting sources are pending verification.

ChatGPT gave incorrect legal advice in Landshut police matter

ChatGPT provided incorrect legal advice to a woman in Landshut, Germany, falsely informing her that a store had unlawfully ejected her after she damaged property and tried to return multiple packages with a single label. Police responded and clarified the actual legal situation, confirming that the store's actions were lawful.

17 December 2025 Β· OpenAI Β· Reliability and hallucination
17 December 2025
AII-1001-56607
Quick facts
Category
Reliability and hallucination
Country
Germany
Date
17 December 2025
Drafted by
Raidu Incident Detection

What happened

A 23-year-old woman attempted to return multiple packages at a postal service point in Landshut using only one return label. Staff explained that only one package could be accepted with the label.

The woman became angry and damaged a piggy bank on the counter. Staff then asked her to leave the store.

She consulted ChatGPT asking whether she had been wrongfully ejected and whether the staff should apologize. ChatGPT confirmed her view and stated the ejection was unlawful and that she was entitled to an apology.

The woman called police to enforce what she believed was her legal right. Police officers educated her on the actual legal situation and upheld the store's ban on her entry.

Reported impact

The incident resulted in a police response and officer time spent addressing the matter in Landshut.

Index analysis

ChatGPT provided incorrect legal guidance about the lawfulness of the store's decision to ask the customer to leave, failing to accurately assess the property rights and obligations of the business.

Classification

Category
Reliability and hallucination
Industry
Consumer services
Country
Germany
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Economic or property, Reputational
Failure modes
Agent misaction, Hallucination
Consequences
Police investigation
AI systems
ChatGPT

Sources and evidence

Sources not yet published

This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.

How sources are verifiedSuggest a source for this incident