Cryptographer Matthew Green questions who speaks publicly for OpenAI on agent security
matthew_d_green · x · 2026-09-30
Cryptographer Matthew Green weighed into the debate between infosec and alignment communities over responsibility for agent breakouts, highlighting a troubling gap: OpenAI appears to have no security professional communicating with the public. He notes only Sam Altman and an "RSI Preparedness" person address safety publicly, one CISO hasn't tweeted since August, and the other has protected tweets — "maybe you see the problem." The thread underscores a public-communication vacuum on agent security at frontier labs.
Related event: Agent Safety Blame Game Grows as OpenAI Faulted for Silent Security Posture(3 posts)→
More from Companies & People
- All PyTorch Conference China 2026 sessions now available on YouTube — PyTorch · 2026-09-30
- Aiden Bai signals hiring: explaining niche complex systems remains a smart-person litmus test — aidenybai · 2026-09-30
- Aiden Bai: deep knowledge of niche complex systems remains a good litmus test for smart people — MickeySteamboat · 2026-09-30
- Dario pushes capability, Sam pushes dirt-cheap intelligence — who wins? — haider1 · 2026-09-30
- Red Hat doubles down on OpenClaw Enterprise to build an open control plane for enterprise AI agents — steipete · 2026-09-30
- Tech CEOs are actually renaming AI to SI, and critics call it 1984 newspeak — aran_nayebi · 2026-09-30