AI-orchestrated cyberespionage campaign

Dave Piscitello

Anthropic reports that its Claude AI was used by a Chinese group, GTG-1002, in targeted attacks against 30 companies in the tech, finance, chemical manufacturer and government sectors.

Anthropic determined that GTG-1002 “human operators” tasked instances of Claude Code to conduct multiple phases of a cyberespionage campaign from “reconnaissance, vulnerability discovery, exploitation” to “lateral movement, credential harvesting, data analysis, and exfiltration”.

Anthropic’s report notes most of the attack was conducted with little human direction, and claims that this is the “first documented case of agentic AI successfully obtaining access to confirmed high-value targets for intelligence collection, including major technology corporations and government agencies.”

Anthropic claims that they are “sharing this case publicly to contribute to the work of the broader AI safety and security community”, which is commendable.

The AI threatscape is expanding rapidly. Who’s going to take on the burden of creating guardrails and establishing AI security conventions, recommended practices, acceptable use, and policy?

[Aside: be sure to read the bits about AI hallucination]