National Vulnerability Database Modernization in the Age of AI

Michael Woolslayer
Policy Counsel
Image
colorful squares connected by lines

The National Vulnerability Database (NVD), run by the National Institute of Standards and Technology (NIST), is a foundational resource that security teams across the public and private sectors use to track software flaws. As artificial intelligence (AI) changes how vulnerabilities are found and exploited, NIST asked for input on modernizing the NVD. HackerOne filed comments on October 13, 2026.

Our central message: AI can scale proven processes and focus human expertise.

Fix the real bottleneck

Finding more vulnerabilities only improves security if organizations can validate and fix them, but that is where systems are now breaking down. On the H1 Platform, the backlog of validated, unresolved vulnerabilities grew 131% in two years, even as average resolution time improved 54% over the same period. 

The answer is not to replace security professionals, but to equip their processes to handle higher volume. AI is well suited to repeatable work such as initial triage, spotting duplicates, and connecting information across reports. That frees people to focus on findings that are novel, ambiguous, or dependent on a specific environment, where their judgment is irreplaceable. Human reviewers should be able to see the underlying evidence and challenge an AI-generated conclusion.

Build on disclosure processes that already work

NIST should not create separate processes for AI-discovered vulnerabilities. Vulnerability disclosure programs, bug bounty programs, and coordinated disclosure already give researchers trusted ways to report flaws, and they should remain the foundation as AI scales discovery. Our role as a launch partner in Patch the Planet shows how existing infrastructure can support AI-enabled discovery at scale, pairing AI-assisted research with expert human review.

NIST should also keep building on established standards: NIST Special Publication 800-216 and the International Organization for Standardization (ISO) standards 29147 and 30111. This is timely given Executive Order 14412, which directs the Federal Acquisition Regulation (FAR) Council to propose rules requiring covered federal contractors to adopt vulnerability disclosure programs consistent with NIST guidelines. That guidance should be practical for the organizations that run these programs.

Prioritize by real-world risk while connecting the ecosystem

As known vulnerabilities pile up, severity scores alone cannot tell teams what to fix first. Prioritization should reflect actual risk in context: how easily a flaw can be exploited, its potential impact, how widespread it is, what mitigations exist, and what attack paths it opens. AI can connect these signals at scale, including spotting low-severity issues that can be chained into a serious exploit. But the final call should rest on the risk a vulnerability poses in practice, not on a score or automated output alone.

That determination depends on data that keeps moving and improving. The NVD should operate as part of the broader vulnerability management ecosystem, not as a standalone repository. NIST should support open, machine-readable mechanisms that carry validated information between the NVD, disclosure programs, vendors, and remediation workflows, so context is preserved and no one re-enters the same details by hand.

NIST should also let vendors, coordinators, and researchers add context or flag inaccurate information, with safeguards for review and clear provenance. Researchers who contribute should get acknowledgement and a visible path to submit corrections. Silence discourages participation, and shrinks the pool of useful information for everyone.

The bottom line

The NVD can remain a trusted anchor for vulnerability management in the AI era, but only if modernization strengthens what already works. AI can help the vulnerability ecosystem operate at greater scale, but it cannot replace human expertise, trusted disclosure processes, and risk-based judgment that helps make vulnerability data reliable. We look forward to working with NIST and the broader community toward that goal.

About the Author

Michael Woolslayer
Michael Woolslayer
Policy Counsel

Michael is Policy Counsel at HackerOne, where he supports public policy efforts to address cybersecurity and AI security challenges and enable good faith security and safety research.