proomt

Search

Search posts, papers, and topics

All posts

Google DevelopersAchuth Narayan Rajagopal4 min readintermediate

Agent Anomaly Detection, now in Private Preview on the Gemini Enterprise Agent Platform

Summary

Google’s Gemini Enterprise Agent Platform now offers a private‑preview Agent Anomaly Detection service that asynchronously analyzes agents’ OpenTelemetry traces and logs to flag risky behavior (e.g., tool misuse, privilege abuse) without adding latency, surfacing findings in Security Command Center and via an API for automated mitigation.

  • Runs out‑of‑band: detection pipeline processes logs and traces asynchronously, preserving agent response latency.
  • Two‑stage analysis: lightweight statistical scan flags sessions, then an LLM‑based reasoning layer inspects flagged sessions for policy violations.
  • Findings include severity, probability, plain‑language rationale, and remediation suggestions; integrated with Security Command Center.
  • API allows programmatic response (e.g., block further tool calls) based on severity thresholds.

As autonomous agents become more capable, risk shifts from code bugs to unexpected runtime behavior. Detecting anomalous tool usage or privilege escalation after the fact is essential for enterprise security and compliance, especially when agents operate without explicit human oversight.

5/10

Related reading

  1. The DevFest Community Workshop Experience: Building Real Agents Together

    Google’s DevFest Community Workshop introduced a “Workbench” format that emphasizes architectural mental models over copy‑paste code, guiding engineers to build long‑running, self‑evolving multi‑agent systems with the Agent Development Kit and Gemini Enterprise platforms. Attendees learned state‑separation, workflow pausing, and self‑patching pipelines, and the series will continue in five more c…

    Google Cloud Bloggoogle.com2 min
  2. Changing the game: How Google uses agentic AI to secure hundreds of millions of lines of code

    Google’s AI & Infrastructure team built an agentic pipeline (Mantis) that runs pre‑submit AI‑driven scans on every code check‑in, validates findings with a fast triage agent (AST + call‑graph analysis) achieving >92% precision in <1 min, then auto‑generates fixes via a bug‑fix agent. Localized threat models and a two‑step scan cut false‑positives to ~3% and prevent hundreds of vulnerabilities eac…

    Google Cloud Bloggoogle.com4 min
  3. Gemini 3.8 Live models now available on AI Gateway

    Vercel AI Gateway now offers Google’s Gemini 3.8 Live and Live Extended Thinking models through its realtime WebSocket API, enabling real‑time spoken interactions with audio output and parallel reasoning. The post includes a short TypeScript example for token acquisition, WebSocket setup, and handling of realtime events.

    Vercelvercel.com1 minrelease