According to BleepingComputer, the GTIG report describes broad AI adoption by financially motivated and state-backed groups. Examples include the use of CAIRN, an open-source multi-agent penetration-testing framework, in an attack on the US aerospace sector, and suspected North Korean hackers using Claude Code to deliver malware. According to Google, attackers systematically abuse open-source pentesting tools, including ARTEX, CAIRN, CyberStrikeAI and HexStrike AI, against internet-exposed servers with low-complexity flaws.
GTIG said that in the past quarter it identified, for the first time, attackers using language models to find and exploit vulnerabilities with no human in the loop. The researchers found an attacker's exposed server hosting a multi-agent system called exploit_agent, which analyzes software updates of popular web apps, looks for signs of flaws and tries to exploit them. The server also held stolen data and hacking tools. The system was used in attacks on French media, entertainment and education websites.
According to Google, these tools sharply cut the time defenders have to patch, and also create noise: after the two PHP2Shell flaws were disclosed, hundreds of exploit demonstrations appeared, mostly AI-generated, and defenders had to check them. The advice for organizations is to patch internet-facing servers quickly. The FBI issued a similar warning about ARTEX last week, as reported here.