As cited
Copy frozen at (site build).
vulnerabilities
Introducing the CyberAgents Exchange AI Inspector: Rigorous review for community-built AI
Tenable and OpenAI announced the CyberAgents Exchange artificial intelligence (AI) Inspector, a security review process for community-submitted artificial intelligence (AI) agents, skills, and related components launching in September. The Inspector combines Tenable's exposure detection, OpenAI's GPT Cyber models, and human oversight to evaluate submissions across multiple risk categories, from prose-only files to weaponized artifacts, with different model tiers applied based on complexity. The review process anchors findings to specific code commits, tests submissions in isolated environments, and works collaboratively with submitters to address security issues before promotion.
Why it matters: Security teams evaluating open-source AI agents need to understand the vetting rigor available before adoption; this Inspector addresses the expanded attack surface of AI systems, including prompt injection, tool-chaining permissions, and instruction obfuscation, that traditional software security review processes do not cover.
- Source published
- First seen by Cybersecurity Tracker