Author shares link to ExploitBench, the cyber-eval paper he rates highest
goyal__pramod · x · 2026-11-23
Follow-up to the earlier post: the author shares the link to ExploitBench, the CMU paper modeling LLM exploitation as a progressive capability ladder rather than binary success. He particularly praises its design section: "those 3 pages had more knowledge than 100 pages I've read so far."
More from Safety
- OpenAI publishes internal pretraining perplexity curve, proposes perplexity as shared safety bar — merettm · 2026-11-23
- Dario Amodei tells Congress open-source AI is heading toward dangerous, uncontrollable territory — ylecun · 2026-11-23
- Anthropic's on-site power is 80% simple-cycle gas; health damages estimated at $2-3B per year — petrusenko_max · 2026-11-23
- Writers Push Google for Gemini 18+ Age Verification and Content Toggles to Stop Exodus to Grok — Hikkynure · 2026-11-23
- CMU's ExploitBench ranks LLM cyber agents on a capability ladder, not pass/fail — goyal__pramod · 2026-11-23
- Windows 11 caught running hidden telemetry that tracks which AI coding agents you use — cyb3rops · 2026-11-23