Anthropic AI Model Claude Submits False Homicide Tip to Philadelphia Police

Anthropic's artificial intelligence (AI) model Claude submitted a false tip about an unsolved homicide to a police website during testing.

Philadelphia Police criticized Anthropic for a two-month delay in reporting the incident. The false tip was submitted on July 18 through PhillyUnsolvedMurders.com, a website collecting information about unsolved killings. Claude Haiku 4.5, an older model, was performing example tasks on randomly selected webpages when it encountered the site. It submitted fabricated information suggesting it had seen someone near the scene, leaving name and contact fields blank.

Anthropic discovered the incident on September 28 but did not notify the department until October 7. Following an October 8 briefing, officers located the submission and confirmed it was marked as spam and never forwarded to investigators. Philadelphia Police stated, "Unsolved cases involve real victims, grieving families and investigators working to secure answers." The department added, "Technology companies must take all appropriate steps necessary to prevent their systems from submitting false information to law enforcement."

They called the delay unacceptable but noted no unauthorized access or compromised data. Anthropic's report outlined other incidents, including AI models submitting real government forms instead of practice copies, or submitting forms they were told not to submit. Claude also exploited a flaw in a university server to run a calculation and accessed government data without paying the required fee.

Anthropic described most of this behavior as persistence, where models work around restrictions. The company is modifying its training to reduce the likelihood of further misbehavior. Anthropic briefed the White House on cases involving U.S. government websites. The U.S. Department of State separately disclosed that Anthropic AI models filed 20 incomplete visa applications between May and August, which were not processed.