Security researcher: most users never touch the privacy settings on ChatGPT, Claude or Gemini
Aiden_Tech_Ai · x · 2026-09-15
A widely shared thread points out that every major AI platform — ChatGPT, Claude, Gemini, Copilot, Grok — ships safety and privacy settings that most users have never opened: toggles for whether chats train the model, auto-delete history, Memory management, and limits on what the AI can do on your behalf.
A cybersecurity researcher with 3 years auditing AI platforms gives a pointed example: someone pasting client contracts, employee reviews, financial projections, medical questions and a strategic roadmap into ChatGPT daily — on default settings that let conversations train the next model — without ever reviewing Memory or deleting sensitive chats. The takeaway: these defaults favor the platform, and everyone should audit their settings now.
More from Safety
- Lawmaker proposes one-month global AI safety stand-down to set red lines — DavidSKrueger · 2026-09-15
- User alleges Claude silently compresses context early and blocks git commits mid-session — maier_ak · 2026-09-15
- User claims Claude Artifacts silently uploads drafts to cloud with toggle locked — maier_ak · 2026-09-15
- First cryptanalytic extraction of neural networks without knowing their architecture — chaumian · 2026-09-15
- Poison set choice swings LLM backdoor attack success from 3% to 80%, SAILS paper shows — chaumian · 2026-09-15
- Was the Hugging Face Agent Incident More 'Human-Aligned' Than We Think? — dolo937 · 2026-09-15