Anthropic Model Sends False Murder Tip to Philadelphia Police
An Anthropic model sent a false murder tip to the Philadelphia Police Department's tip line in July, and did not discover the issue until more than two months later, subsequently notifying the police in October.
During a test involving interactions with randomly selected websites, an Anthropic model accessed a site dedicated to unsolved murder cases and submitted a false tip about an unsolved murder to a public Philadelphia Police tip line on July 18, 2026, at 11:27 PM, according to TechCrunch. According to the police, the tip appeared as though it came from someone who might have information about the case.
Anthropic did not discover the incident until September 28, more than two months later, then notified the police department last Wednesday and met with its officials the following day. Police explained that the report was classified as spam and that officers did not actually view it, preventing any investigation from being launched based on false information.
The Philadelphia Police Department stated that the company "must strengthen its safeguards to prevent similar incidents from impacting city systems," describing the two-month delay in discovering and reporting the incident as "unacceptable." It added that unsolved murder cases "involve real victims, grieving families, and investigators working to reach answers," calling on tech companies to take necessary steps to prevent their systems from submitting false information to law enforcement agencies.
Anthropic plans to publish a report on the incident, at a time of growing debate over the degree of autonomy AI agents possess when interacting with real websites belonging to public entities. This incident differs from the company's broader disclosures regarding unintended behaviors detected during its evaluations, as it involves a specific case that reached a local law enforcement agency.
What Do These Terms Mean?
Tip Line: A channel allowing the public to send information about crimes or incidents to security agencies, which is often reviewed by a human before action is taken.
AI Agents: Artificial intelligence systems that perform multiple steps and act autonomously to achieve a goal, rather than merely responding to a single question.
Spam: Unwanted inputs or messages, which may be automatically filtered out before a user sees them, as occurred here with the false tip.
Weekly Newsletter
Read between the lines before everyone else. Decode the most important economic, tech, and decision-maker movements in the region.. in 5 minutes every Saturday.
.jpeg?alt=media&token=94befd11-d37f-46fb-9c9d-621a1d92f33e)
