Meta patched a security flaw in its Meta AI chatbot that allowed users to access other users' private prompts and responses. Security researcher Sandeep Hodkasia discovered the bug, which involved easily guessable prompt numbers, and received a $10,000 bug bounty for responsibly disclosing it. Meta confirmed the fix in January 2025, stating no evidence of malicious exploitation. The vulnerability highlighted security challenges in the rapidly developing AI landscape.

Prepared by Jonathan Pierce and reviewed by editorial team.