Federal Agencies Must Deploy AI Cyber Defenses Within 30 Days as CVE Disclosures Surge to Historic Levels
U.S. federal agencies face a June 2026 deadline to harden systems with AI-enabled cyber defenses as annual CVE projections soar to 66,000.
New Federal Mandate on AI-Enabled Cyber Defenses
Federal agencies must begin hardening their information systems with AI-enabled cyber defenses within 30 days of June 2, 2026, according to a presidential directive from The White House.
In tandem with this requirement, the Secretary of the Treasury, in consultation with the National Cyber Director, the Secretary of War, and the Secretary of Homeland Security, shall form an AI cybersecurity clearinghouse within 30 days to coordinate scanning for software vulnerabilities, discover and validate vulnerabilities, and coordinate remediation and distribution of vulnerability patches.
CVE Disclosures Reach Historic Highs
The timing of this mandate comes as the cybersecurity landscape faces unprecedented pressure. The updated 2026 CVE forecast projects approximately 66,000 CVEs for the full year, up from the February median of 59,427, marking the first time in history that annual vulnerability disclosures are on pace to approach 70,000, according to FIRST (Forum of Incident Response and Security Teams).
Actual CVE disclosures through April 2026 are running 46.3% above the projections published in February 2026, with 6,420 excess CVEs recorded.
Structural Drivers of the Surge
Three structural factors are fueling the 2026 CVE explosion. AI-assisted vulnerability discovery is accelerating disclosure rates. GitHub Security Advisory (GHSA) volume has surged 449% year-over-year. Additionally, VulnCheck CNA-of-Last-Resort activity has increased 3,119%.
A notable example: there was a 164% spike in Q1 CVE disclosures from the Mozilla CNA, directly attributable to AI-assisted tooling running against the Firefox engine.
The number of distinct software products with tracked vulnerabilities has also grown by two orders of magnitude, driving workload independent of AI or CNA changes.
Actionable Risk Remains Stable
Despite the surge in raw CVE volume, when filtered for real-world risk—using CISA KEV entries or EPSS scores above 10%—actionable exploitability remains flat and the patching burden has not materially increased.
Source: The White House
Developments since publication
-
Within 60 days of the Executive Order, agencies shall develop and maintain a classified benchmarking process to assess the advanced cyber capabilities of AI models and determine the threshold at which Source
-
On December 4, 2025, a 17-year-old was arrested in Osaka under Japan's Unauthorized Access Prohibition Act for running malicious code to extract personal data of over 7 million users of Kaikatsu Club, Source
-
In February 2025, three teenagers aged 14, 15, and 16 with no coding background used ChatGPT to build a tool that hit Rakuten Mobile's system approximately 220,000 times. Source
-
In July 2025, a single actor using Claude Code conducted an extortion campaign targeting 17 organizations over one month, using agentic AI to develop malicious code, organize stolen files, analyze fin Source
-
In December 2025, an individual used Claude Code and ChatGPT to breach the Mexican government, targeting more than 10 agencies and stealing over 195 million taxpayer records. Source
-
Malicious packages discovered on public repositories increased by 75% in 2025. Source
-
Cloud intrusions increased by 35% in 2025. Source
-
Time to exploit dropped from over 700 days in 2020 to 44 days in 2025. Source
-
Mandiant's M-Trends 2026 report found that 28.3% of CVEs are exploited within 24 hours of disclosure, meaning time-to-exploit has effectively gone negative. Source
-
Top frontier models including ChatGPT, Claude, and Gemini improved performance on SWE-bench from 33% in August 2024 to just under 81% by December 2025. Source
-
There were 55,000 malicious packages in public repositories in 2022, which grew to 454,600 by 2025 according to Sonatype. Source
-
In September 2025, the Shai-Hulud attack compromised over 500 npm packages, leading to 487 organizations having secrets compromised. Source
-
An amateur in Algeria built ransomware using AI that hit 85 targets in his first month. Source