---
format: "aidr-story-markdown/v1"
id: "548af7c4772d0fa2901b6bebac9dc16fad928ac02fa6faa031793ba4070af229"
canonical_url: "https://aidr.today/548af7c4?lang=en"
title: "How Agent Anomaly Detection scores on benchmarks we didn't write"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-11-12T19:39:48.000Z"
category: "Tools"
topics: ["google","safety","devtools"]
source_urls: ["https://developers.googleblog.com/how-agent-anomaly-detection-scores-on-benchmarks-we-didnt-write/"]
summary: "Agent Anomaly Detection (AAD) provides asynchronous, low-latency monitoring for AI agents by analyzing OpenTelemetry traces and logs to flag goal hijacks and tool misuse. Tested across five public security benchmarks, AAD successfully caught 87% of harmful instructions carried out by agents without adding runtime latency."
---

# How Agent Anomaly Detection scores on benchmarks we didn't write

> [Open the canonical story](<https://aidr.today/548af7c4?lang=en>)

**Published:** 2026-11-12T19:39:48.000Z
**Category:** Tools
**Topics:** google, safety, devtools

## Summary

Agent Anomaly Detection \(AAD\) provides asynchronous, low\-latency monitoring for AI agents by analyzing OpenTelemetry traces and logs to flag goal hijacks and tool misuse\. Tested across five public security benchmarks, AAD successfully caught 87% of harmful instructions carried out by agents without adding runtime latency\.

## Sources

- [Story source](<https://developers.googleblog.com/how-agent-anomaly-detection-scores-on-benchmarks-we-didnt-write/>)