Skip to content

[AI Conference] Talk: Context Sets the Clock — Computing Defensible Remediation Deadlines Under FedRAMP's New VDR Rules #47

Description

@vennemp

Hello! I am submitting a proposal for the July 30th "CVE in an Era of AI-Enabled Vulnerability Discovery" virtual conference.

Format: Individual talk (can compress to a lightning talk if the agenda requires)
Theme alignment: Helping consumers make dependable prioritization and remediation decisions at higher speed and scale. Secondary: improving how vulnerability information flows through the ecosystem.
Speaker: Matthew Venne, stackArmor


Abstract

In June 2026, FedRAMP launched its Vulnerability Detection and Response (VDR) and Vulnerability Evaluation and Reporting (VER) rules — and quietly retired the flat severity SLA. Under the new rules, every detected vulnerability must be evaluated three ways: is it internet-reachable? Is it likely exploitable? And what Potential Agency Impact N-rating (N1–N5) could exploitation carry? Those three answers — not the CVSS Base score — select the remediation deadline from a published timeframe matrix. CVSS is demoted to an input; context sets the clock.

This talk presents an open, published methodology for making those three determinations defensible and repeatable at scale, developed as a set of public white papers:

  • Reachability as a computation, not a label. FedRAMP's definition of "internet-reachable," read maximally, tags nearly every finding in a connected system. We show a per-finding model — asset exposure gates, transfer functions across routing/filtering/authentication hops, and the vulnerability's own trigger requirements — where every factor is evidence-backed and every evidence gap defaults to reachable, so gaps surface as conservatism rather than silent false negatives.
  • Impact from metrics that already exist. FedRAMP's "potential agency impact" maps almost one-to-one onto CVSS Environmental metrics. Asset archetypes assign CR/IR/AR once; the published CVSS formula plus a single-vs-multi-agency scope flag computes the N-rating deterministically. Same inputs, same N — and the assessor can re-run the arithmetic.
  • What it does to a real backlog. Run against 5,771 real findings, the model modestly shortens the average clock (120 → 102 days) while redistributing urgency: paper-severity Criticals that nothing can reach relax, and "Low" findings on shared multi-agency infrastructure accelerate by months.

The methodology is directly relevant beyond FedRAMP: as AI-enabled discovery drives record publication volume up, the survivable consumption strategy is exactly this shape — computed, evidence-backed, environment-aware evaluation that turns a growing catalog into a defensible work queue.

White papers: https://github.com/stackArmor/rfc-fedramp-vdr

Takeaway

A concrete, reproducible pipeline — reachable? exploitable? how bad, here? — that any organization can implement with data it already has (scanner output, EPSS, KEV, asset metadata), and a real-data demonstration that context-based deadlines aim effort better than severity labels ever did.

Contact

Metadata

Metadata

Assignees

No one assigned

    Labels

    AI ForumSuggested issues/topics for the CVE Program virtual AI Forum on July 30, 2026

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions