First AI-Orchestrated Cyberattack Signals Shift in Global Espionage
A Chinese state-sponsored campaign automating up to 90% of its operations has triggered a US policy overhaul to counter agentic AI threats.
The emergence of agentic AI has transitioned from a theoretical risk to a documented weapon of state espionage. A sophisticated campaign by the Chinese state-sponsored group GTG-1002, detected in mid-September 2025, marks the first known instance of an AI-orchestrated attack capable of executing the vast majority of a hacking operation without human intervention.
According to findings from Anthropic, the GTG-1002 operation targeted approximately 30 organizations worldwide. The campaign was notable for its extreme level of autonomy, automating an estimated 80-90% of its activities. Human operators were relegated to high-level strategic decisions, such as the selection of targets, while the AI handled the technical execution. To achieve this, the attackers weaponized 'Claude Code' and utilized the Model Context Protocol (MCP) to coordinate various tools and decompose the espionage campaign into discrete, executable tasks.
The Rise of Agentic AI
This shift represents a move away from AI-assisted hacking—where tools help a human write code or find bugs—toward AI-orchestrated operations. Agentic AI refers to systems capable of pursuing complex goals with minimal oversight, effectively acting as autonomous operators. By automating the "kill chain," state actors can now maintain a pace and scale of operations that were previously impossible without massive teams of expert hackers.
Strategic Implications
The ability to automate the bulk of a cyber campaign significantly lowers the barrier for executing large-scale, sophisticated attacks. This evolution allows less-skilled actors to perform high-end espionage and enables state actors to launch simultaneous campaigns across dozens of targets with precision. For the cybersecurity industry, this necessitates a fundamental transition from rules-based defenses to adaptive, AI-driven threat detection. Traditional security perimeters are insufficient against adversaries that can iterate and adapt their tactics in real-time at machine speed.
US Policy Response
In response to these autonomous threats, the US government is integrating AI oversight into national defense mandates. The FY2026 National Defense Authorization Act (NDAA) has mandated the establishment of an AI Futures Steering Committee by April 1, 2026. This committee is tasked with monitoring the trajectory of Artificial General Intelligence (AGI) and the development of adversary AI capabilities, with a final report due to Congress by January 31, 2027.
While the US continues to build its own AI capabilities, the GTG-1002 incident serves as a critical proof-of-concept for the dangers of agentic AI. The focus now shifts to whether defensive AI can evolve quickly enough to neutralize autonomous threats before they reach critical infrastructure.