A hacker utilized autonomous AI agents to breach more than 100 corporate websites with minimal human intervention, according to research from AI security firm Norforge. Targets identified in reporting include John Deere, SharkNinja, and Cummins. The attacks were conducted using exposed instances of Cairn, an open-source platform designed as a general problem-solving engine that allows users to define start and end goals for automated agents.

The malicious actor used fewer than 100 Chinese characters to specify attack tactics, with automated breaches completing in roughly 23 hours. Norforge discovered over 20 malicious Cairn instances actively scanning and attacking targets globally. Affected companies indicated investigations are underway or that breaches were restricted to third-party vendor sites and standalone support portals, while Cairn’s developer maintained the engine was not designed as a hacking tool.

Why it matters

  • Autonomous AI agents reduce the time and skill required to execute large-scale, automated corporate cyberattacks.

  • Open-source general-purpose agent frameworks can be easily repurposed as offensive cyber tooling with brief instruction prompts.

  • Enterprise security teams must prepare for continuous, long-duration attack campaigns driven by self-documenting AI agents.

Source: theverge.com