Dev48
Language
  • About
  • Services
  • Industries
  • Technologies
  • Articles
  • Contacts
Book a call
    Home/Articles/Nvd modernization vulnerability prioritization in the ai era
Dev48
info@dev48.org+375 29 320-40-96WhatsApp@isky_by

© 2026 · All rights reserved.

NVD Modernization: Vulnerability Prioritization in the AI Era

Source: HackerOne•November 18, 2026 at 04:10 PM

NVD Modernization: Vulnerability Prioritization in the AI Era

HackerOne’s NIST comments on NVD modernization explain why vulnerability prioritization must reflect real-world risk. Read the full response.

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

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.

← All articles

What this article says

More in Cybersecurity

All →
CVE-2026-123007: Critical NGINX HTTP/2 Integer Underflow Leads to RCE (FIXED)
Rapid7

CVE-2026-123007: Critical NGINX HTTP/2 Integer Underflow Leads to RCE (FIXED)

Why your software needs a CBOM
SonarSource

Why your software needs a CBOM

Unauthenticated Remote Code Execution in NGINX: CVE-2026-123633 and CVE-2026-123007
Wiz

Unauthenticated Remote Code Execution in NGINX: CVE-2026-123633 and CVE-2026-123007

Open by Default, Already Occupied: Threat Actors are Renting your Compute
Rapid7

Open by Default, Already Occupied: Threat Actors are Renting your Compute

Cognizant and Qualys Redefine AI-Era Managed Security Service
Qualys

Cognizant and Qualys Redefine AI-Era Managed Security Service

ESET builds on the Windows endpoint security platform to advance endpoint resilience
ESET

ESET builds on the Windows endpoint security platform to advance endpoint resilience

More from HackerOne

Continuous Security Validation: Close the Testing Gap
HackerOne

Continuous Security Validation: Close the Testing Gap

AI Security Accountability: Who Owns the Call?
HackerOne

AI Security Accountability: Who Owns the Call?

Measuring the Fix: Frontier Models and CTEM Remediation
HackerOne

Measuring the Fix: Frontier Models and CTEM Remediation

Critical PHP Vulnerabilities: What Security Teams Need to Know
HackerOne

Critical PHP Vulnerabilities: What Security Teams Need to Know

The Exposure Debt Benchmark
HackerOne

The Exposure Debt Benchmark

What’s New for HackerOne Researchers
HackerOne

What’s New for HackerOne Researchers