Pentagon reports tenfold rise in cybersecurity vulnerabilities due to AI

US Army Lt. Gen. Paul T. Stanton, director of the Defense Information Systems Agency, said at the Billington CyberSecurity Summit on September 10 that AI has driven a tenfold increase in exploitable zero-day vulnerabilities across Pentagon computer systems. Stanton attributed the rise to roughly 30 years of deferred maintenance on Defense Department networks, where AI can now scan massive codebases and chain individually low-severity flaws into full system compromises. The Pentagon is responding with an Army cyber command AI task force and Project Griffin, a program building defensive AI agents to detect and respond to automated attacks.

Pentagon reports tenfold rise in cybersecurity vulnerabilities due to AI Decades of deferred network maintenance have left US defense systems dangerously exposed as AI hands adversaries powerful new tools to exploit forgotten flaws The US military has spent nearly three decades pouring money into next-generation weapons systems. Meanwhile, the computer networks those weapons depend on have been quietly rotting from the inside out. Now, artificial intelligence is turning that neglect into an active crisis. At the Billington CyberSecurity Summit on September 10, US Army Lt. Gen. Paul T. Stanton disclosed that AI has driven a tenfold increase in exploitable zero-day vulnerabilities across Pentagon computer systems. Stanton, who serves as director of the Defense Information Systems Agency, framed the problem in blunt terms: deferred maintenance on Defense Department networks poses potential peril to national security. Three decades of digital debt The core issue isn’t that the Pentagon’s networks were poorly designed. It’s that they haven’t been properly maintained in roughly 30 years. During that stretch, funding priorities consistently favored shiny new weapons platforms over the mundane but essential work of patching, upgrading, and hardening the IT infrastructure underneath. AI technologies have dramatically lowered the barrier for adversarial access to these aging systems. Minor software flaws that would have been individually harmless, the kind of bugs that sit in backlogs for years because no single one seems urgent, can now be identified and chained together by AI into potent zero-day exploits. AI, tech, and the markets they move—in one daily briefing. Daily. Free. Join 34,000+ readers across crypto, finance, and policy. The democratization of elite hacking Stanton’s warning centers on a troubling shift in the threat landscape. AI agents can now perform tasks that were previously limited to a small cadre of highly skilled hackers. The technology doesn’t just speed up existing attack methods. It fundamentally expands who can execute them. The tenfold increase in vulnerabilities isn’t because the Pentagon’s software suddenly got ten times worse. What changed is that AI can now scan massive codebases, identify subtle flaws humans would miss, and automatically determine how to combine those flaws into attack chains. A vulnerability that a human analyst might classify as low-severity becomes a critical threat when an AI figures out it can be paired with three other low-severity bugs to achieve full system compromise. Playing defense with the same technology The Pentagon isn’t sitting still. Stanton outlined several initiatives aimed at fighting AI with AI. An Army cyber command AI task force is working to integrate machine learning into defensive operations, and a program called Project Griffin is focused on building defensive AI agents capable of detecting and responding to the same kinds of automated attacks that adversaries are now deploying. Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy https://cryptobriefing.com/editorial-policy/ .