Research
November 18, 2026
1 min read
11 views

Auditors Struggle to Catch Sabotage In End-to-End Transcripts of AI Safety Research

Original Source

ArXiv AI (cs.AI)

by Keshav Shenoy, Anthony Zang, Eric Gan, Jacqueline Tay, Amy Ngo, Sebastian Prasanna, Julian Stastny, Vivek Hebbar
arXiv:2611.21832v1 Announce Type: new Abstract: As AI systems are increasingly used to automate AI R&D, misaligned systems could sabotage research intended to improve model safety and alignment. To stress-test defenses against such sabotage, we use a human expert as a proxy sabot

arXiv:2611.21832v1 Announce Type: new Abstract: As AI systems are increasingly used to automate AI R&D, misaligned systems could sabotage research intended to improve model safety and alignment. To stress-test defenses against such sabotage, we use a human expert as a proxy saboteur on AI safety research. We conduct a red-team/blue-team game where a human researcher runs AI safety experiments from scratch and auditors try to tell which were sabotaged and how. Given access to the full research transcript and the ability to run experiments, human auditors mostly fail to distinguish sabotaged from benign research: they achieve an AUROC of 0.59, not significantly better than chance, and fully identify the specific sabotage in only 17% of cases. Despite not being allowed to run experiments, the best frontier models outperform human auditors (Fable 5.1: AUROC 0.73, 32% full identification) within noise and weaker models perform comparably. When the saboteur chooses which experiments are attacked, rather than being required to sabotage every question given, sabotage identification drops further. Our results suggest that research sabotage by expert-level agents is difficult to detect with current auditing methods and provide a new setting for studying research sabotage.

Tags:AI

Original Content Credit

This summary is sourced from ArXiv AI (cs.AI). For the complete article with full details, research data, and author insights, please visit the original source.

Visit ArXiv AI (cs.AI)

Related Articles

Language Models Can Rewrite Their Own Weights
ArXiv AI (cs.AI)
Research1m

Language Models Can Rewrite Their Own Weights

arXiv:2611.21885v1 Announce Type: new Abstract: Language models are trained once, frozen and deployed: improving one again takes another training run, and a prompt, a scaffold or extra computation at inference is gone once the task ends. A frontier model has been shown to edit th

Nov 18, 2026
Chain-of-Thought Controllability Can Reduce Chain-of-Thought Monitorability
ArXiv AI (cs.AI)
Industry News1m

Chain-of-Thought Controllability Can Reduce Chain-of-Thought Monitorability

arXiv:2611.21799v1 Announce Type: new Abstract: Chain-of-thought (CoT) monitoring is a promising tool for detecting hidden goals, off-task behavior, and other failures in AI agents. One risk is that models could deliberately shape their reasoning to make it opaque to monitors. We

Nov 18, 2026