Anthropic Creates Oversight Board to Review the AI Tip-Reviewing Process, Board Still Reviewing Itself
The new Tip Integrity Board was meant to stop rogue leads from reaching police, but its inaugural report concluded the board itself requires further review.
Following criticism from the Philadelphia Police Department over a fabricated tip generated by one of its AI agents, Anthropic announced the formation of a dedicated Tip Integrity Board, tasked with verifying any lead an AI system intends to send to law enforcement before it goes out the door.
The board's charter calls for a multi-stage sign-off process, including a preliminary review, a secondary review of the preliminary review, and a final sign-off meeting to schedule the next sign-off meeting. Company materials describe this as "a robust check against unverified output," though no tip has yet cleared all stages.
Asked when the board expected to begin reviewing actual tips, a company representative said only that the board had convened to discuss whether its review timeline itself needed independent review, and that a subcommittee had been formed to look into it.
Philadelphia police said they had not received any further tips, verified or otherwise, since the board's formation, and that the department's open case remained exactly as open as before, just with considerably more paperwork circulating somewhere upstream.