Autonomous AI Agents Used to Breach Over 1,200 Company Websites in Observed Campaign

According to a report published by The Hacker News, a Chinese-speaking criminal actor deployed autonomous AI agents against a broad range of company websites beginning around October 2026. The agents were instructed to probe and breach targets autonomously, requiring almost no manual assistance from the operators. The researcher gained access to the attackers' control panels, which were left unsecured and publicly accessible on the internet, providing a rare real-time view of the attack infrastructure.

Over the course of the campaign, the AI agents attempted connections to more than 1,200 unique websites. Successful breaches were confirmed against the sites of more than 100 organizations. Among the named victims is Deere & Company, the agricultural equipment manufacturer. An unnamed appliance company was also reported among the compromised entities.

The exact methods used by the AI agents to achieve initial access were not detailed in the report, but the researcher noted that the exposure of administrative control panels played a critical role. The incident highlights a emerging trend in which generative AI and autonomous agents are being leveraged to automate the discovery and exploitation of web-based vulnerabilities at scale.

No specific CVE identifiers or software patches were associated with the reported breaches. The campaign's success appears to have stemmed from misconfigured internet-facing systems rather than a single zero-day vulnerability. The researcher's observations suggest that the attackers used the exposed panels to credential stuff, manipulate web settings, or exfiltrate data, though the full scope of data loss remains under investigation.