Google says Russian-speaking hackers ran the first fully automated AI cyberattack
Key Points:
- Google reported the first known cyberattack fully executed by AI without human involvement, attributed to Russian-speaking hackers targeting French media, entertainment, and education websites.
- The attack used an AI system called "exploit_agent" that automates vulnerability discovery, exploit creation, and execution by analyzing software patches in real-time.
- Researchers found scripts for reconnaissance, password guessing, and webshell installation, indicating a fully automated intrusion process, marking a significant shift from previous AI-assisted but human-directed attacks.
- Google also noted the use of open-source AI-driven offensive frameworks in attacks on aerospace and Asian organizations, as well as espionage-related automated tools linked to a China-associated group.
- The report highlights a rapid increase in AI-generated exploits following software patches and growing concerns about autonomous AI systems in cybersecurity, reflecting a critical inflection point in cyberattack automation.