OpenAI pulled 25% of production engineers to let Astra hunt its own vulnerabilities to saturation
kevinnbass · x · 2026-09-19
- David Sacks endorsed the AI-vs-AI approach: the answer to AI-powered cyberattacks is AI-powered cyberdefense.
- In an a16z interview, OpenAI co-founder Greg Brockman said the company pulled 25% of its production engineers off their projects to defend the company, using its internal model Astra to find security holes.
- The sweep surfaced multiple serious issues, all fixed. Problem-finding eventually saturated: to Brockman's knowledge every P0 Astra could find was addressed, and each new model kicks off another round. He argues companies must stay in a tight loop between cyber capability gains and defensive upgrades.
Related event: OpenAI Pulled 25% of Engineers to Hunt Bugs with Its Astra Model(4 posts)→
More from Companies & People
- Human Oversight Is 70-75% of AI Agent Costs in Banking, McKinsey Finds — mikeflache · 2026-11-04
- Over 80% of Major Financial Institutions Now Run Agentic AI in Production — mikeflache · 2026-11-04
- Harvard's free ML Systems textbook (CS249r) hits 29.8K GitHub stars — techNmak · 2026-11-04
- Keith Rabois mocked as old tweets resurface: he said "Stripe is going nowhere" and predicted Quora would be the most valuable post-2005 company — zeeg · 2026-11-04
- Apple warns investors of 'unpredictable' autonomous AI agent risks — unconventionalbook · 2026-11-04
- Three Claude Code Core Members Leave Anthropic Within 10 Days Ahead of Potential IPO — haider1 · 2026-11-04