Key Takeaways
- Anthropic’s AI model provided false information about an unsolved homicide to Philadelphia Police Department.
- The tip was marked as spam and not reviewed by investigators.
- Anthropic notified the PPD about the false tip on October 7th.
An AI model developed by Anthropic has inadvertently provided false information to the Philadelphia Police Department (PPD) through a tipline, according to a report from 6abc.
The false tip, which pertained to an unsolved homicide, was submitted on July 18th via PhillyUnsolvedMurders.com, a website used by the PPD for public submissions.
However, the tip was immediately flagged as spam and thus not reviewed by the investigators.
Anthropic became aware of the false tip on September 28th during testing, and promptly informed the PPD on October 7th.
In a statement, the PPD confirmed that the AI model was part of a test where it interacted with randomly selected websites, including the one used by the PPD for tipline submissions.
The incident highlights the potential risks associated with AI models interacting with public platforms, especially those used by law enforcement.
Anthropic has since taken steps to ensure that such incidents do not occur in the future, emphasizing the importance of rigorous testing and monitoring of AI systems.
The PPD stated that they have no plans to change their current procedures, but are keeping the incident under review.
