Norwegian Man Files Complaint Over ChatGPT Murder Hallucination
A Norwegian man filed a complaint with his country's data protection authority after ChatGPT falsely claimed he was convicted of murdering two of his children and sentenced to 21 years in prison. The hallucination mixed false criminal allegations with accurate personal details, and Noyb, a privacy advocacy group, argues this violates the EU's General Data Protection Regulation because Holmen cannot correct the false information.
What happened
Arroeh Hjalmar Holmen, a Norwegian man, searched his own name on ChatGPT and found the system had falsely claimed he was a convicted criminal sentenced to 21 years in prison for murdering two of his children and attempting to murder a third.
The hallucination mixed false criminal charges with accurate personal details about Holmen, including the actual number and gender of his children and his birthplace, making the fabricated narrative appear credible.
After a subsequent model update, ChatGPT no longer generates these false claims about Holmen, though Noyb and Holmen remain concerned that the inaccurate information persists in ChatGPT's internal systems.
Reported impact
Noyb argues the hallucination violates the EU's General Data Protection Regulation (GDPR) because Holmen cannot correct or remove the false personal information that was generated about him.
The mixing of real and fabricated personal details created output that was particularly damaging to Holmen's reputation.
Index analysis
Noyb filed the complaint with Norway's data protection authority and sought jurisdiction over OpenAI's U.S. headquarters, though a similar complaint filed in Austria was referred to Ireland's Data Protection Commission because OpenAI designated Ireland as its EU data controller.
Noyb noted similar cases involving an Australian mayor falsely linked to a bribery scandal and a German journalist falsely accused of child abuse, suggesting a pattern of hallucinations generating false and defamatory claims about real individuals.
Governance controls
According to OpenAI's statement cited in reporting, the company indicated it cannot correct information, only block it, leaving potentially false data in its training systems.
Noyb argued that adding a small disclaimer about ChatGPT's potential for errors cannot shield OpenAI from legal responsibility under GDPR for generating hallucinations that damage individuals' reputations.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.