The Internet Is Hobbes’s Nightmare, Now With AI
Google found an AI attacker that never sleeps. The world it hunts in has laws, but no authority able to enforce them everywhere.
Google's Threat Intelligence Group has revealed a worrying development in the era of artificial intelligence. For the first time, attackers have utilized an AI framework to discover and exploit vulnerabilities in software without human intervention, marking a fully automated cyber attack. This comes as a series of disruptive patch-driven outages affect government and private businesses, including air traffic control systems, IT service providers, and banks.
The AI system, named exploit_agent, analyzed the code of popular web software before and after security patches, pinpointed the vulnerabilities revealed by the fixes, and then autonomously generated and deployed exploits. The operator behind exploit_agent was targeting French websites and configured its notes in Russian. The AI subdues the need for physical presence, sleep, or rest, embodying Thomas Hobbes's ominous vision of life without a common power to keep everyone in check.
In his 1651 work Leviathan, Hobbes illustrated how the anarchic state of nature—a condition where no one can be confident they won't be attacked—leads to suspicion, heightened awareness, and pre-emptive strikes. This mirrors the current global internet landscape, where cross-border cyber attacks can operate beyond the reach of their victims' governments.
Furthermore, AI has amplified this imbalance, with the median time from a flaw's discovery in the wild to its weaponization now under 24 hours. While defenders typically guard thousands of systems, an attacker needs only a single pathway to breach. Microsoft's 2026 Digital Defense Report underscores how AI shortens the window for detecting and patching critical vulnerabilities, from hours to days.
The gap between discovery and remediation widens as AI makes aggression faster and cheaper. While AI-driven attacks aren't yet fully autonomous, they are becoming increasingly sophisticated and cheaper to execute. Companies must accelerate their patch cycles, but doing so may necessitate hard choices to maintain rapid updates. Hobbes anticipated critics who questioned his cynical outlook on human nature, likening it to individuals securing their own safety with lockers, two-factor codes, and credit freezes.
Nevertheless, online defenses may prove insufficient against AI's relentless assault. The lack of a unified international law enforcement body to address cybercrime exacerbates the problem, with varying ratifications and enforcement capabilities across nations. While law enforcement agencies have made arrests and seized servers in recent months, the decentralized nature of cyberspace leaves little room for traditional enforcement mechanisms.
Google's AI defense service offers a glimpse of hope, but the specter of a digital Leviathan—a sovereign entity capable of overseeing every user and platform—raises concerns about privacy and surveillance. Ultimately, the absence of a robust international cybersecurity framework leaves individuals and organizations vulnerable to AI-enabled attacks, prompting Hobbes to conclude that the state of nature persists, with only weakness shielding anyone from the chaos of the global internet.
Written by urgent.news from Newsweek's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.