Particle.news
Download on the App Store

FBI Flag of ChatGPT Threat Leads to RAID Arrest in Strasbourg With No Charges Filed

The case shows how AI safety checks can route threats to French police through Pharos.

Overview

  • A 37-year-old from Strasbourg was arrested by the RAID on Friday, April 3, after he asked ChatGPT how to get a gun to kill an intelligence agent from the CIA, Mossad, or France’s DGSI.
  • U.S. investigators at the FBI detected the exchange and alerted French authorities via Pharos, France’s portal for reporting illegal online content.
  • Police searched the man’s home and found no weapon, and prosecutors later dropped the case because the conduct was limited to questions posed to an AI.
  • The Strasbourg prosecutor said the man had prior psychiatric issues, and he was involuntarily hospitalized after his custody ended.
  • OpenAI states that in very rare cases of serious and imminent danger, AI conversations can be escalated to human review or law enforcement, a policy that helps explain how this alert reached police.