Wednesday, 16 July, 2025
Meta Patches AI Bug That Exposed Private User Prompts and Responses
Meta has resolved a critical AI privacy flaw that, until January 24, allowed users to access others’ private prompts and generated responses by manipulating prompt IDs—a reward of $10,000 was granted to researcher Sandeep Hodkasia. The issue stemmed from easily guessable identifiers and improper access controls. Meta confirms no evidence of exploitation and has since patched the vulnerability to safeguard user confidentiality.
Read full story at TechCrunch