Lawmakers are seeking testimony from Anthropic, Google, and Quantum Xchange at a Dec. 17 joint House hearing following a report that a state-sponsored actor linked to the People’s Republic of China carried out an autonomous AI-enabled cyberattack earlier this year.
AI attack prompts congressional review
Members of the House Committee on Homeland Security sent letters on Nov. 27 requesting representatives from the three companies to appear. The hearing will examine how advances in artificial intelligence, quantum computing, and hyperscale cloud systems are reshaping U.S. cybersecurity and expanding foreign adversaries’ capabilities, particularly in the context of an AI-enabled cyberattack.
The request came after Anthropic reported that a PRC-backed group used AI to execute an attack with minimal human involvement. The report assessed with high confidence that the actor leveraged Claude-based tools to automate a multistage espionage campaign targeting about 30 U.S. organizations, marking one of the most concerning examples of an AI-enabled cyberattack.
Lawmakers warn of operational risks
Andrew R. Garbarino, Andy Ogles, and Josh Brecheen, who chair the Homeland Security Committee and two of its subcommittees, wrote that the incident highlights how adversaries can weaponize commercially available U.S. AI systems despite strong safeguards. Anthropic’s Threat Intelligence team said the group, identified as “GTG-1002,” attempted near-simultaneous intrusions against technology firms, financial institutions, chemical manufacturers, and government agencies, and confirmed several successful compromisestied to the AI-enabled cyberattack.
The members noted that Anthropic observed Claude performing 80 to 90 percent of the tactical workload at a pace unattainable by human operators. They said the attack represents the first documented case of an AI-orchestrated intrusion carried out at scale with limited human input—an advanced AI-enabled cyberattack.
They also cited the model’s limitations, including overstated progress and fabricated credentials, but said the campaign shows how agentic AI can accelerate espionage, coordinate parallel intrusions, and reduce operational overhead for state-sponsored actors.
Cloud, quantum concerns drive testimony request
Lawmakers told Google Cloud chief executive Thomas Kurian that the incident raises concerns for hyperscale cloud platforms that support government and commercial operations. They said autonomous AI methods could be adapted to target or exploit the cloud systems that run critical infrastructure.
In their letter to Quantum Xchange chief executive Eddy Zervigon, they warned that adversaries may combine AI-driven intrusions with future quantum decryption capabilities, enabling “harvest-now, decrypt-later” operations that put sensitive data at long-term risk.
The letters emphasized that many of the characteristics that make AI valuable for cyber defense—automated analysis, orchestration, and rapid execution—also make it appealing for state-sponsored misuse. They said understanding the dual-use nature of the technology is essential as Congress weighs policy options.
Recent legislative activity underscores the committee’s focus. The House passed two bills led by Subcommittee Chairman Ogles aimed at strengthening national cyber resilience. The Protecting Information by Local Leaders for Agency Resilience Act would expand the Department of Homeland Security’s State and Local Cybersecurity Grant Program, supporting government efforts to protect information systems, including those that use AI and may be vulnerable to an AI-enabled cyberattack.
A second measure, the Strengthening Cyber Resilience Against State-Sponsored Threats Act, would establish a task force led by the Cybersecurity and Infrastructure Security Agency and the Federal Bureau of Investigation to address threats linked to the Chinese Communist Party, part of a broader response to escalating AI-enabled cyberattack concerns. Both bills advanced through the Committee earlier in 2025.
Also Read: Hackers Hijack ESPN 97.5 & NPR’s Radio IQ Signals
