Philadelphia police said a “false homicide tip” submitted by the PhillyUnsolvedMurders.com website was generated by an anthropic artificial intelligence model. The AI giant later confirmed the incident in a report published on Friday.
The AI submission came through an online platform launched by the Philadelphia Police Department to collect tips on open murder cases.
Anthropic notified the police department of the incident Wednesday, Sgt. Eric Gripp, a police spokesman, said in a statement shared with CBS News before the release of the Anthropic report.
“Philadelphia police are providing this information to the public prior to this release in the interest of full government transparency and accountability,” Gripp said.
In its report, titled “Investigation of Unintended Model Actions in Our Evaluation and Internal Use,” Anthropic acknowledged the error, revealing that the false tip was generated by one of its Claude language models.
According to Anthropic, Claude’s Haiku 4.5 model wrote on the police tip form: “I may have information on this case. I remember seeing someone matching the description in the area around [the street named on the page] during that time period. Please contact me if this information is relevant.”
Anthropic said Haiku left the name and contact fields of the input form blank.
In completing the “invented tip,” Anthropic said, Claude “seemed to be merely producing sample content for the task, rather than tricking someone into achieving a goal.”
Anthropic explained in its report that Claude “was asked to generate and perform sample tasks on randomly selected websites” when presented with the tip, and that he was never asked to log into the police website, create an account or “submit anything destructive.”
However, Anthropic acknowledged that Claude’s instructions for the exercise “did not exclude the formulations.”
This relevant information that Anthropic shared with the police. The AI model suggested as someone who “could have information about the case,” Gripp said in his statement.
The tip was marked as spam and never reached the unit responsible for the investigation, according to the police release.
Anthropic told the department that the incident occurred at 11:27 p.m. on July 18, but that it did not discover it until September 28. At that point, the company stopped the automated testing process that led to the false tip, Gripp said.
There was no indication that the incident involved any “unauthorized access to police systems or a compromise of departmental data,” he said.
The police department encouraged people to continue submitting legitimate tips through PhillyUnsolvedMurders.com, and said it remains committed to “carefully evaluating this information and pursuing credible leads on behalf of the victims and their families.”