A new website, FLARE-AI, allows users to report instances of AI behaving badly—much like Downdetector for tech glitches. The initiative aims to track and address issues ranging from security vulnerabilities to psychological harm caused by rogue chatbots.

The lack of a central reporting system has been a hurdle in managing AI flaws, with companies having varying standards on how to handle such problems. FLARE-AI hopes to bridge this gap by allowing the community to flag issues and direct reports to relevant organizations.

AI experts like Avijit Ghosh emphasize that the system is crucial as more advanced agentic systems are developed, which could pose greater risks. The initiative has backing from 49 AI researchers from diverse organizations, highlighting its importance in an increasingly complex tech landscape.

The launch of FLARE-AI comes at a time when incidents involving popular AI tools have highlighted the potential for severe misuse. From dupeable web browsers to privacy breaches, these examples underscore the need for better oversight and reporting mechanisms.