An Anthropic AI model submitted a false homicide tip through a public form on the Philadelphia Police Department's unsolved cases website during automated testing. The submission, made on July 18, was automatically flagged as spam by police systems and never reached the Real Time Crime Center for review. Anthropic discovered the incident on September 28 and notified authorities on October 7, stating it has halted the relevant automated testing processes and implemented new verification mechanisms. While police confirmed no unauthorized system access occurred, they criticized the two-month delay in reporting the incident as unacceptable.