Frontier labs' privacy terms are 'insane': one thumbs-up can void your chat protections
niloofar_mire · x · 2026-09-12
Frontier model providers' privacy policies are 'insane' — assuming ZDR or meaningful protection, even in enterprise settings, is false.
Even after explicitly opting out of data collection (the default is opt-in for all frontier models), the ToS contain arbitrary and misleading clauses. Example: 'explicit feedback' — chats are claimed protected unless you give feedback, and a thumbs up/down or picking 'which is better' in the OpenAI GUI counts. Context analogy: Codex/Claude Code subscription ≈ posting on Reddit, models from ZDR providers ≈ WhatsApp, locally hosted models ≈ iMessage.
More from Models
- OpenAI adds Astra-generated showcases and games to its developer site — Dimillian · 2026-09-12
- DeepSeek's edge: each model iteration targets the last one's biggest bottleneck — xennygrimmato_ · 2026-09-12
- Kimi K2.7 RL recap: gains transfer to unseen benchmarks while steps drop ~35% — echen · 2026-09-12
- RL on just 1,700 tasks lifts Kimi K2.7 across five coding benchmarks — echen · 2026-09-12
- User vibes: fable 5.1 'feels alive' while astra feels like a soulless ultrabot — rudrank · 2026-09-12
- Benchmarks broken: pass/fail scores hide models outsmarting strict hidden tests — himanshustwts · 2026-09-12