AI companies establish divergent policies on threat reporting to authorities
AI companies adopt opposite threat-reporting policies: Anthropic blocks surveillance use; ChatGPT enables threat disclosure to authorities.
Event / Security
A 37-year-old man from Strasbourg was arrested after making threats about acquiring weapons during conversations with an AI system. The FBI detected his statements and reported them to French authorities via the PHAROS platform. He was released without charge after no weapons were found and was committed to mandatory hospitalization. The case highlights broader debates about AI use in surveillance and threat reporting policies.
2026-08-19 to 2026-11-16 · UTC publication dates
Coverage begins 6 Apr, 09:36 UTC
Follow the documented timeline or explore the related reporting.
Anthropic, a California-based AI firm, recently refused to allow its artificial intelligence system Claude to be used for mass surveillance of US citizens under the Trump administration. This refusal provoked a strong reaction from the US Department of Defense, which subsequently turned to OpenAI, the creator of ChatGPT, for AI capabilities.
A 37-year-old man from Strasbourg was arrested on Friday after making statements during a conversation with an artificial intelligence application about his intention to purchase weapons to kill intelligence agents from the CIA, Mossad, or DGSI. His statements were intercepted by FBI investigators, who reported them to the French PHAROS platform for reporting illicit online content and behavior. The man was arrested by RAID officers on Friday afternoon.
The man arrested in Strasbourg had his detention lifted on Saturday, and the case was closed without further prosecution. No weapons were found at his residence during the search. The suspect, who has a history of psychiatric care, was committed to mandatory hospitalization.
ChatGPT stated that in very rare cases of grave and imminent danger, statements submitted to the AI system can be transmitted to human teams and potentially to authorities according to applicable laws. OpenAI clarified that such reporting is rare and targeted in nature, not automatic.
AI companies adopt opposite threat-reporting policies: Anthropic blocks surveillance use; ChatGPT enables threat disclosure to authorities.
Strasbourg man arrested for weapon threats via AI, then released without charges.