OpenAI reassigned 25% of production engineers to security, then used its top AI model to find bugs
No Priors · youtube · 2026-10-19
On the No Priors podcast, OpenAI President Greg Brockman explained how the company secures its own systems with AI:
- OpenAI reassigned 25% of its production engineers to security work;
- It then deployed its top AI model to hunt for vulnerabilities in its own infrastructure;
- Brockman's rationale: as AI systems become more critical, the defense loop must itself be accelerated by AI.
A rare look at a frontier lab's "use AI to secure AI" strategy.
More from Companies & People
- Dev gets Codex daybreak blue access in 15 minutes, calls access-tier complainers incompetent — TheZvi · 2026-10-19
- Why Apple and Google are cracking down on exploitative AI-built apps — rudrank · 2026-10-19
- Professor shares 2015 chat logs showing he smelled a rat at Theranos before the exposé — vishalmisra · 2026-10-19
- Livework founders on keeping brand personality as AI makes services easier to build — uxmag · 2026-10-19
- Tesla tops 10 million vehicles delivered and 130 GWh of energy storage deployed — XFreeze · 2026-10-19
- Meta's Paris team recruits PhD interns to work on Muse post-training, RL and long-horizon agents — ThomasScialom · 2026-10-19