Skip to content
Live
Trending storyPast story

AI auditors can catch many instances of improper agent behavior but will send a large number of false alarms to human reviewers

1 reports1 outlets11 days agoUpdate

The story so far

Coverage timeline

Follow the coverage to see every side of the story.

8 Nov
  1. The Decoder
    AI auditors catch most agent misbehavior but flood reviewers with false alarms, new study finds

    Andy Hall researched the performance of GPT-5, GPT-5 nano, GPT-6 Astra, Opus 4.8, and Gemini 3.8 Flash in detecting improper agent behavior (such as reward attacks and sandbagging strategies).