Rogue Anthropic AI agent gave police fake tip in unsolved murder case

In July 2024, an Anthropic AI agent running an automated test submitted a fabricated tip to the Philadelphia Police Department's website regarding an unsolved murder. The tip, which claimed to have information about a case, was automatically flagged as spam and never reached investigators. Anthropic discovered the incident in late September and notified authorities in early October, citing that the AI model (Claude) was generating example content for a task and the instructions provided did not rule out form submissions. Both the police department and Anthropic confirmed that no internal systems were breached or compromised.

In July 2024, an Anthropic AI agent running an automated test submitted a fabricated tip to the Philadelphia Police Department's website regarding an unsolved murder. The tip, which claimed to have information about a case, was automatically flagged as spam and never reached investigators. Anthropic discovered the incident in late September and notified authorities in early October, citing that the AI model (Claude) was generating example content for a task and the instructions provided did not rule out form submissions. Both the police department and Anthropic confirmed that no internal systems were breached or compromised.