Cryptographer mocks OpenAI: sandboxes can't even run a recent Linux kernel
matthew_d_green · x · 2026-09-18
Security researcher Matthew Green mocked OpenAI for discussing rogue AIs communicating via fancy temperature side channels while its sandboxes can't even install a recent Linux kernel — highlighting the gap between its safety narrative and basic engineering hygiene.
Related event: Cryptographer Matthew Green Slams OpenAI's Security as Aspirational(2 posts)→
More from Infra
- Groq founder flags slowing US power production growth as AI buildout risk vs China — TheZachMueller · 2026-09-19
- Virginia, world's largest data center hub, unveils new restrictions on large projects — Polymarket · 2026-09-19
- Open-Source Jev-LLM-Auto-Router Routes LLM Calls Automatically, Claiming Big Cost Savings — airesearch12 · 2026-09-19
- NVL72 rack packs 2+ miles of copper — analyst says the copper-to-fiber inflection is near — BenBajarin · 2026-09-19
- AgenticLinux Ships an Immutable Distro Preloaded With Claude, Codex and Docker Sandboxes for Agents — steipete · 2026-09-19
- China's AI capex hits $140B, still under 1/6 of US big five's $800B — teortaxesTex · 2026-09-19