Overview
- On July 31, OpenAI’s CISO Dane Stuckey announced the shutdown of ChatGPT’s “discoverable” chat sharing feature, which was disabled the following morning.
- OpenAI is collaborating with Google and other search engines to scrub previously indexed conversations, although cached results may persist beyond deletion.
- Investigations found shared chats contained job applicant evaluations, mental health disclosures, crime confessions and proprietary code, highlighting unintended data exposures.
- Meta AI’s Discover feed remains fully public and continues to be indexed by Google, and the company has no plans to block search access to shared chats.
- The episode underscores the importance of clear consent settings and robust privacy safeguards in generative AI platforms.