OpenAI shifted 5-10% of its compute from training to safety work, says CRO Mark Chen
pstAsiatech · x · 2026-10-04
OpenAI chief research officer Mark Chen says that over the past couple of months the company has moved between 5% and 10% of its vast compute resources away from training new models and toward safety work, especially monitoring. In an interview he also covers how the firm makes its models safe, how a deliberate slowdown would work, and why he believes the world is better off with OpenAI in it, adding they won't "shoot ourselves in the foot" over hack fallout. The poster notes Chinese AI developers would struggle to do anything comparable.
More from Companies & People
- ElevenLabs' startup grant program drew tens of thousands of startups and now drives 10% of enterprise revenue — thedealdirector · 2026-10-04
- OpenAI Employee Recaps First DevDay, Moderating Closing Q&A With Sam Altman — paw_lean · 2026-10-04
- Weekly AI roundup: White House puts Gemini and Grok on 29,000 federal sites, Google Beam ships — shashib · 2026-10-04
- Blogger: Public ideas beat resumes — traditional CVs pigeonhole cross-domain thinkers — signulll · 2026-10-04
- Aave DAO can't own its trademark, so Aave Labs proposes a memberless Cayman foundation — LexSokolin · 2026-10-04
- Granola cofounder on designing products when agents skip your UI for MCP — petergyang · 2026-10-04