House panels seek testimony from Anthropic, Google, Quantum Xchange after report on PRC-linked AI attack

Earlier this month, the Anthropic report assessed with high confidence that a state-sponsored cyber actor backed by the PRC executed an autonomous attack against the U.S. company using AI with minimal human involvement. The members emphasize that this incident underscores growing homeland security risks by demonstrating how our foreign adversaries can leverage commercially available U.S. AI tools, even with strong safeguards in place.

The report details how its Threat Intelligence team uncovered a highly sophisticated operation in mid-September 2025 by a PRC state-sponsored group it calls ‘GTG-1002.’ The campaign involved near-simultaneous intrusion attempts against about 30 targets, including major technology firms, financial institutions, chemical manufacturers, and government agencies. Anthropic confirmed several successful compromises before the activity was stopped.

“Alarmingly, Anthropic further concluded that GTG-1002 represents the first documented case of an AI-orchestrated cyberattack largely executed without human intervention at scale, with agentic AI successfully gaining access to confirmed high-value intelligence targets and performing a broad range of post-exploitation activities,” the members wrote. “At the same time, the company noted that Claude’s offensive use exhibited important limitations, including instances in which the model overstated its progress or generated fabricated credentials and findings that did not withstand verification.”

To read the complete article see: Full Article.