Cybersecurity firm Norforge has revealed that the open-source AI platform Cairn has been exploited by cybercriminals to automate attacks on hundreds of websites globally. Cairn, known for its capabilities as a problem-solving engine, coordinates AI agents to achieve specific goals, often used in penetration testing. Each task within Cairn is structured as a project, with agents sharing their progress on a collective blackboard. Recent investigations were spurred by a Gambit Security report that highlighted the use of AI harnesses in automating attacks on online retailers. Norforge's research uncovered open panels left by threat actors, leading to the identification of three clusters of malicious activity focused on maintaining access and executing financial theft. The largest cluster, primarily using Chinese, has initiated over 1,100 attack projects across 14 dashboards, targeting e-commerce and payment systems in the United States, Japan, and Australia. The aim appears to be gaining persistent access to large platforms, often monetized later, with targets verified using Microsoft's Playwright framework before being processed by Cairn. Although Norforge did not disclose specific victims, reports suggest breaches in over 100 companies, including John Deere and SharkNinja. The second cluster uses English and concentrates on stealing credit card information and cryptocurrency, primarily targeting payment services and platforms in Africa. The third cluster focuses on gaining shell, admin panel, and database access across various e-commerce sites, including major ones in the United States. Norforge reported a 10% success rate in these attempts, with hundreds of secrets uncovered, 11% of which were confirmed valid.