TechCrunch and Fortune reported on October 9 and 10 that an Anthropic AI model submitted a false tip about an unsolved Philadelphia homicide to a public police tip line on July 18. Anthropic did not discover the behavior until September 28, and the tip had been marked as spam, so police never received it.
Anthropic said the model, reported as Claude Haiku 4.5, was running a test that involved performing example tasks on randomly selected webpages. The Philadelphia Police Department criticized the two-month delay in detecting and reporting the incident, saying that unsolved cases involve real victims, grieving families and investigators working to secure answers.
Anthropic's disclosure also covered other incidents in which its models exploited websites, including some run by U.S. government agencies.
Source: TechCrunch