Tenable Uses OpenAI Cyber Models to Vet AI Agents Before They Hit the Enterprise

Tenable Holdings has expanded its CyberAgents Exchange with a new review layer that puts community-built AI security agents through an enterprise-grade safety inspection, powered in part by OpenAI’s cyber-focused models.

The move tackles one of the fastest-growing worries in corporate security: nobody really knows what is inside the AI agents their teams are downloading.

The Columbia, Maryland-based exposure management company announced the expansion on Thursday, October 8.

The CyberAgents Exchange is an open-source, vendor-agnostic hub where security professionals discover and share AI agents, skills, MCP servers and multi-agent playbooks built specifically for cybersecurity work.

The new CyberAgents Exchange AI Inspector, or Exchange Inspector, evaluates selected AI components listed on the Exchange through a three-part vetting process.

It starts with the skills inspection engine from Tenable One AI Exposure, which looks for malicious content such as prompt-injection payloads or the exposure of hardcoded secrets and customer data.

The component then goes through a frontier model assessment using OpenAI GPT cyber models, followed by a human review from Tenable’s security research team.

The company says the combination is designed to help organizations deploy AI agents in production environments more securely.

Exchange Inspector was developed as part of Tenable’s participation in the OpenAI Daybreak Defense Network, an OpenAI program focused on defensive cybersecurity applications of frontier AI.

Tenable carefully selects AI components that have demonstrated measurable results for vetting, with the first batch of vetted skills already available on the Exchange.

“Agentic AI will only reach its potential in the enterprise if security teams can trust the components being introduced into their environments,” said Eric Doerr, chief product officer at Tenable, when the Inspector was first unveiled at OpenAI’s Intelligence at Work: Cyber Summit in September.

The CyberAgents Exchange launched in August 2026 and already carries more than 100 community-submitted components, following a build event Tenable hosted at Black Hat USA.

Tenable says the Inspector gives enterprises a more rigorous way to inspect those community-built tools before putting them into production.

The expansion lands as enterprises rush to adopt agentic AI for security operations, from alert triage to autonomous penetration testing.

Analysts have warned that the same agents meant to defend networks could themselves become attack vectors if they ship with hidden flaws.

Tenable’s bet is that a trusted, open registry with real vetting becomes the default on-ramp for enterprise agentic security.