Skip to content

Back to the catalog

Atlasのご紹介：WizのAI脆弱性リサーチャー
wiz.ioarticle

Atlasのご紹介：WizのAI脆弱性リサーチャー

Wiz Research details Atlas, an autonomous multi-agent AI system for vulnerability discovery that ranks #1 on the CyberGym benchmark (90.9% success) and found 200+ previously unknown bugs in heavily audited projects like grpc, dnsmasq, Kubernetes, gVisor, the Linux kernel, and containerd. The post explains Atlas's four-stage pipeline (attack-surface mapping via code property graphs, parallel hypothesis hunting, adversarial validation between competing agents, and dynamic exploit triggering) and notes an early version of the system contributed to discovering a critical GitHub RCE (CVE-2026-3854) that paid GitHub's largest-ever bug bounty. It also outlines lessons from evaluation-driven development and plans to integrate Atlas into Wiz Code alongside Google CodeMender.

TL;DR Wiz's Atlas is an agentic AI vulnerability-hunting system that tops the CyberGym benchmark and found 200+ real bugs (plus a GitHub RCE, CVE-2026-3854) using code-property-graph mapping and adversarial multi-agent validation.
Open original ↗

Organisations named here

  • Wiz Research

CVEs named here

  • CVE-2026-3854

Technologies named here

  • Atlas
  • grpc
  • dnsmasq
  • CyberGym
  • Kubernetes
  • Cyber Model Arena
  • gVisor
  • Google CodeMender
  • Linux kernel
  • Wiz Code
  • containerd
  • Wiz Security Graph
  • GitHub
  • Red and Green Agents
  • code property graph
  • Wiz

Techniques named here

  • Attack surface mapping
  • Parallel hypothesis hunting
  • Adversarial validation
  • Dynamic exploit triggering
  • Evaluation-Driven Development

Related in AI Security