KEV Catalog (Known Exploited Vulnerabilities)
The KEV Catalog, short for Known Exploited Vulnerabilities Catalog, is a list published by the US Cybersecurity and Infrastructure Security Agency (CISA) of CVEs with confirmed evidence of active exploitation. A vulnerability enters the catalog only when CISA has reliable proof that attackers are using it against real targets. Under Binding Operational Directive 22-01, US federal civilian agencies must remediate KEV entries by a set due date, and most private security teams treat the catalog as a prioritized do-this-first patch list. For AI systems, a KEV entry affecting a model server or MCP dependency moves to the front of the queue.
How the KEV Catalog is built
CISA adds a CVE to the catalog when three conditions hold: the vulnerability has an assigned CVE ID, there is reliable evidence of active exploitation in the wild, and a clear remediation action exists (usually a vendor patch). Each entry carries a due date. Binding Operational Directive 22-01 gives federal civilian agencies a fixed window to remediate, often two to three weeks, which is why the catalog functions as an exploitation-confirmed shortlist rather than a theoretical severity ranking.
Why the KEV Catalog matters for AI infrastructure
An AI deployment inherits a long dependency tree: inference servers, orchestration frameworks, retrieval components, and MCP servers, each pulling in open-source libraries. When one of those components appears in the KEV Catalog, the exploitation is no longer hypothetical. Pairing the KEV Catalog with EPSS scoring gives a security team a plain rule: patch KEV entries in the AI request path first, then work down the EPSS-ranked remainder. DeepInspect records which identities routed traffic through each endpoint, which tells responders whether an exploited component actually sat in a live AI path.
Related reading
- EPSS and KEV Patch Prioritization for the AI Stack: A Working Guide
CVSS severity alone tells you a vulnerability is bad, not whether it is being exploited. This guide shows how to combine the FIRST EPSS exploitation probability score with the CISA Known Exploited Vulnerabilities catalog to rank patch work across your AI inference stack, with a scoring rule, a worked example on real 2026 AI-tool CVEs, and an honest line on what patch prioritization does and does not cover.
- AI Bug-Hunting Drove ~1,500 High-Severity CVEs in June 2026: Prioritizing the Patch Window
AI models are now finding software flaws at scale. Epoch AI counted roughly 1,500 high-severity and critical CVEs reported in June 2026, about 3.5x the previous monthly record, while real exploitation still arrives within hours of disclosure. This walks through why patch prioritization by exploitability, plus identity-bound inline policy at the AI request boundary, is the control that holds between disclosure and patch.