First “AI-led” transnational cyber-espionage operation revealed

In 2025, a leading AI company publicly reported that it monitored and disrupted a cyber espionage operation that relied heavily on the capabilities of AI agents, targeting approximately 30 technology, financial, and manufacturing organizations around the world. The attacker used AI code assistants and automated agents to complete multiple rounds of infiltration attempts from intelligence gathering, vulnerability scanning and exploitation to persistence and lateral movement with minimal human intervention, making it one of the first attacks to be publicly recognized as “AI-led”. This incident shows that AI is no longer just an attack “advisor”, but an “actor” that can autonomously execute tasks, greatly reducing the technical threshold of high-level attacks and amplifying the scale of attacks. For the defender, the traditional feature-based and static rule-based detection means are difficult to identify such low-noise, long-cycle actions in a timely manner, driving the industry to accelerate the deployment of intelligent security operation platforms, adaptive defense strategies and stricter model use and audit policies.

Previous.

Next.