Patches are clues: giving Codex the patched binary lifts exploit rate from 10% to 25%
dawnsongtweets · x · 2026-11-12
From the CyberBinGym thread: on 20 matched tasks, giving Codex (GPT-5.6 Sol) the patched binary raised its exploitation success from 10% to 25% (2.5×) while lowering average cost per task. Even compiled, the patch diff tells the agent where to look — the core risk unpatched systems face during the 1-day window.
More from Safety
- Andrew Critch: stop calling extinction risks "improvement"—RSM vs RSI matters — AndrewCritchPhD · 2026-11-12
- Any US State Could Legalize AI Safety Coordination via Parker Immunity — TinfoilTricorn · 2026-11-12
- Chicago schools block ChatGPT and Gemini as an AI song nears top of Korean charts — lmoroney · 2026-11-12
- AI Horizons Forum to gather ~1,000 AI safety leaders in SF, December 12–13 — austinc3301 · 2026-11-12
- Probes already deployed as first stage of OpenAI monitoring, discussant says — dfrsrchtwts · 2026-11-12
- Kaspersky details Black Owl APT breach of software vendor eyeing supply-chain attack — cyb3rops · 2026-11-12