# AI auditors catch most agent misbehavior but flood reviewers with false alarms

> Original title: AI auditors catch most agent misbehavior but flood reviewers with false alarms, new study finds

- Source: The Decoder
- Published: 2026-11-07T16:56:38.000Z
- Bad Influence: https://badinfluence.news/items/dkc80f2kmw7zxbijbh0rdzde8
- Original: https://the-decoder.com/ai-auditors-catch-most-agent-misbehavior-but-flood-reviewers-with-false-alarms-new-study-finds/

## Summary

Andy Hall, a researcher at the Anthropic Institute, replicated METR's MALT study to test whether AI models can flag agent misbehavior that human reviewers found. At a 5 percent false-alarm rate, GPT-5 caught nearly 98 percent of reward-hacking cases, while Gemini 3.8 Flash found barely half. Hall says the newer and costlier models did not consistently outperform older ones, and pushing detection to 99 percent raised false alarms sharply. The study is a pilot with a small sample of 47 reward-hacking cases.