AI Cyberattacks Outpace Defenses as Global Rules Lag
Google reported the first fully automated cyber attack, where AI found and exploited software flaws without human involvement, intensifying concerns about AI-driven threats to global internet security. The AI system, named "exploit_agent," analyzed software patches to identify vulnerabilities and launched attacks autonomously. Google counted 22 exploited zero-days in August alone, while Microsoft noted the median time from flaw discovery to weaponization is now under 24 hours, far outpacing typical patch cycles of 30 to 60 days. The report highlights a Hobbesian "state of nature" online, with no global authority to enforce laws across borders. International treaties like the Budapest Convention and U.N. Cybercrime Convention remain fragmented, while experts call for faster defenses and enforceable cross-border rules.