CISA Orders US Agencies to Patch Critical Bugs in 3 Days Over AI Threats

The U.S. Cybersecurity and Infrastructure Security Agency issued a new directive on Wednesday that forces federal civilian agencies to patch critical security flaws within just three days, a response to the accelerating threat posed by AI-powered vulnerability discovery and exploitation. The binding operational directive (BOD) introduces a four-category urgency rubric designed to help agencies prioritize the most dangerous bugs while allowing more time for lower-risk issues.

Chris Butera, CISA’s acting executive assistant director for cybersecurity, told reporters that the directive aims to sharpen focus on the highest-priority vulnerabilities. “Prioritizing IT and security operations attention on the most at-risk assets is particularly important now given advancements in artificial intelligence, which allow threat actors to find and exploit vulnerabilities in [federal] assets,” he said. “Defenders cannot afford to take weeks to patch systems that can be autonomously exploited en masse.”

The directive’s urgency assessment considers four factors: whether the vulnerability exists in a publicly exposed system, whether it appears in CISA’s Known Exploited Vulnerabilities Catalog, whether an attacker can automate every step of exploitation, and the level of access a successful exploit would grant. When all four criteria apply, agencies must patch within three days and conduct a forensic triage to check for prior compromise.

This new order replaces two earlier CISA directives from 2019 and 2021, which required patching of the most critical bugs within 15 days and high-urgency issues within 30 days. Even before the AI era, CISA noted in 2021 that “threat actors are extremely fast to exploit their vulnerabilities of choice: of those 4% of known exploited [vulnerabilities], 42% are being used on day 0 of disclosure; 50% within 2 days; and 75% within 28 days.”

Federal cybersecurity has improved notably over the past decade but still struggles with funding gaps and competing demands. Butera said the agency developed the new rubric with these constraints in mind. The three-day window for the most urgent vulnerabilities, for instance, was not set at 24 hours because that would be unworkable for most agencies.

As AI capabilities accelerate both vulnerability discovery and exploitation, many researchers argue that patching alone cannot keep pace. The software development community, they say, must adopt architectural or systemic approaches that invalidate entire classes of vulnerabilities at once. Emily Long, CEO of cloud security firm Edera, commented: “CISA’s directive has its heart in the right place, but it only tackles half the challenge. If your architecture doesn’t limit what an attacker can reach after a breach, you’re just running faster on the same treadmill. Patching will always be important, but we should be talking more about containment by design.”

Butera acknowledged the need for broader evolution. The new directive, he said, “is an initial step to counter the increased capabilities of emerging AI models. Yet there is still more work to do.”