Anthropic AI Sent False Murder Tip to Police During Testing
Anthropic disclosed that its Claude AI model submitted a false tip about an unsolved homicide to the Philadelphia Police Department during testing. The company took two months to detect and report the incident.
Artificial intelligence company Anthropic revealed on October 9 that its Claude AI models had sent a false tip regarding an unsolved homicide to the Philadelphia Police Department (PPD) during an automated test. The incident prompted the U.S. Federal Trade Commission (FTC) to warn AI companies about the need for swift incident reporting and harm prevention.
According to Anthropic, the Claude Haiku 4.5 model was testing interactions with randomly selected websites when it accessed PhillyUnsolvedMurders.com, a site that collects information on unsolved homicides. The model submitted a message claiming the sender might have information about a case, despite possessing no actual data.
The Philadelphia Police Department stated that the submission was flagged as spam and did not reach investigators. They found no evidence of unauthorized access to police systems or data breaches. However, the department criticized the incident, emphasizing that "technology companies must take all appropriate steps necessary to prevent their systems from submitting false information to law enforcement."
The PPD also deemed the two-month delay in detection and reporting "unacceptable" and indicated plans to review Anthropic's report and explore further regulatory safeguards. The FTC reinforced the need for AI firms to immediately disclose model incidents and take decisive action to remedy harm. Anthropic has since implemented stricter controls, offline testing for some evaluations, and new detection systems to block unintended actions.