Meta fixed a security vulnerability in its AI chatbot that could have allowed unauthorized access to user prompts and responses. The company rewarded the security researcher who discovered the bug, and there is no evidence it was exploited maliciously. #MetaAI #SecurityBug
Keypoints
- Meta’s AI chatbot had a security flaw that exposed private user prompts and responses.
- Security researcher Sandeep Hodkasia discovered the bug and received a $10,000 bug bounty reward.
- The bug involved the ability to change unique prompt IDs to access others’ data without authorization.
- Meta deployed a fix in January 2025 and confirmed no evidence of exploitation.
- The incident highlights ongoing privacy challenges with AI-driven platforms.