You Can Now Sound the Alarm on AI Behaving Badly

A new crowdsourced platform called FLARE-AI has been launched to report and track harmful AI behaviors such as generating malware or leaking personal data. Developed by AI researchers including Avijit Ghosh from HuggingFace, it aims to centralize AI flaw reporting and improve transparency. The system routes verified reports to model makers and organizations like MITRE, addressing the current lack of coordinated AI harm disclosure.