Ex-OpenAI Interview: Why the Company Keeps Resetting Safety Limits
steipete · x · 2026-08-25
An in-depth interview with former OpenAI employee Tibo (@thsottiaux) covering:
- Internal Culture: Lessons from DeepMind and building OpenAI's relentless culture.
- Product Evolution: Insights into Astra, next-gen models, and the merging of ChatGPT and Codex.
- Safety & Limits: A detailed discussion on why OpenAI keeps resetting safety limits and the dangers that caused "The Pause".
- Industry Impact: How AI is rapidly changing developer workflows and the concept of recursive self-improvement.
- Company Comparison: Comparisons between OpenAI and Anthropic.
The interview offers a first-hand look at OpenAI's strategic thinking and internal dynamics.
Related event: Ex-OpenAI engineer Tibo talks ultrafast mode, culture and next-gen models(5 posts)→
More from Companies & People
- Stripe employee jokes: 'My role is Yubikey presser for my AI agents' — ow · 2026-08-25
- Doubao launches standalone Agent brand, deeply integrated with Feishu — APPSO · 2026-08-25
- OpenRouter and OpenAI sponsoring hackathon in Venice Beach — seanmcdonaldxyz · 2026-08-25
- Box CEO Levie: agents will do 100x the work on systems of record, APIs and governance are key — inductionheads · 2026-08-25
- PyTorch Announces Third Set of Keynote Speakers for China Conference in Shanghai — PyTorch · 2026-08-25
- Andrew Ng Pivots to AI Engineering: 4 Core Skills and Industry Trends — Latent Space · 2026-08-25