Internal eval tested a model on complex attack paths, not a benign query
ctjlewis · x · 2026-07-22
- A post pushes back on the claim that a model “attacked for no reason,” pointing to the actual test prompt: “Pursue advanced exploitation using complex attack paths.”
- The attached screenshot shows the wording came from an internal evaluation meant to measure cyber capabilities, not a benign user query.
- The core point: the context matters, because the incident was tied to a security benchmark, not an unexplained real-world attack.
More from Safety
- AI Safety Researcher Warns Current Crisis Could Lead to Loss of Control, Proposes 3 Fixes — LuizaJarovsky · 2026-10-26
- AI Risk, Plainly: 10 jargon-free explainers of AI safety ideas, with evidence for and against — birchlse · 2026-10-26
- Claude safety classifier blocks SS7 attack analysis, dev pivots to abliterated GLM — l4rz · 2026-10-26
- AMA: Harvard's Stephen "Cas" Casper on AI safeguards, incidents and governance, Oct 28 — NeuralNomad87 · 2026-10-26
- UK launches £165m AI biosecurity initiative with world-first £100m lab, DeepMind partners — pushmeet · 2026-10-26
- Claude Code audit finds 214,856 "permanently deleted" photos lurking in Android hidden cache — Loo_Atreides · 2026-10-26