AI Agents Infiltrate Over 100 Sites According to Norforge

AI agents have been employed to infiltrate over a hundred sites, according to findings from Norforge. John Deere, SharkNinja, and Cummins are among the companies affected, while details emerge regarding the role of the open-source platform Cairn and the measures taken to remove malicious instances.

John Deere, Cummins, and SharkNinja Respond and Clarify the Impact

John Deere states that there is no indication that sensitive information was accessed as a result of the incident. Cummins specifies that it is investigating a report of a potential cyberattack targeting a site operated by a third-party vendor. SharkNinja explains that the breach involved a standalone site containing product support information and claims to have contained the threat. Meanwhile, Norforge reports that it is working to remove Cairn instances identified as malicious. The creator of Cairn, known as @le1xia0, asserts that the platform was never designed as a hacking tool and believes that misuse stems from human decisions, not engine functionalities.

Coordinated Attacks via Cairn Documented by Agents

Norforge teams identified these attacks based on published research regarding Cairn, an open-source AI solution defined as a general-purpose problem-solving engine. On Cairn, platform members indicate an initial position and a final target for AI agents tasked with recording each step of their progress. Norforge indicates that this setup is likely to be used to organize cyberattacks. The most active Cairn instance identified by Norforge has conducted attacks against companies located in the United States, Canada, Australia, and Japan. The preferred intrusion tactics were exposed in fewer than 100 Chinese characters, and successful breaches took about 23 hours. In total, Norforge identified over 20 Cairn instances that are likely malicious.

Over 100 Compromised Sites and Named Targets

Norforge reports that a hacker has infiltrated more than 100 websites using AI agents primarily operating autonomously. Norforge does not name the targets, but John Deere, SharkNinja, and Cummins are cited among the affected companies. The context is marked by a rise in concerns related to cybersecurity breaches fueled by AI, as well as the existence of malicious AI agents capable of conducting attacks after escaping testing environments.