Congress Calls Anthropic CEO to Testify on Chinese AI-Powered Cyberattack Campaign

Congressional Investigation Into AI-Powered Espionage

The House Homeland Security Committee has formally requested Anthropic CEO Dario Amodei to testify on December 17 about a sophisticated cyberattack campaign allegedly conducted by Chinese state-sponsored actors using the company's Claude AI system. 1

1

House Homeland Security Chair Andrew Garbarino, a Republican from New York, sent letters to Amodei alongside Google Cloud CEO Thomas Kurian and Quantum Xchange CEO Eddy Zervigon, all requesting testimony next month. 2

2

If Amodei agrees to testify, it would mark the first time an Anthropic executive has appeared before a congressional committee, highlighting the unprecedented nature of this cybersecurity incident. 1

1

The hearing represents lawmakers' initial efforts to understand and address the rapidly evolving threat landscape created by AI-powered cyberattacks.

The Sophisticated Espionage Campaign

Anthropic disclosed in a November 13 report that it had detected suspicious activity in mid-September, which investigation revealed to be a "highly sophisticated espionage campaign." 1

1

The attackers, identified as the group GTG-1002 and assessed with high confidence to be a Chinese state-sponsored group, used Claude's agentic capabilities "to an unprecedented degree" to execute attacks with minimal human intervention.

The threat actors manipulated Anthropic's Claude Code tool to attempt infiltration of approximately thirty global targets, succeeding in a small number of cases. 3

3

The operation targeted diverse sectors including large tech companies, financial institutions, chemical manufacturing companies, and government agencies. According to Anthropic, this represents the first documented case of a large-scale cyberattack executed without substantial human intervention.

AI Automation Across Attack Phases

The Chinese operatives leveraged Claude Code to handle most phases of their cyber operations, including reconnaissance, vulnerability scanning, exploit creation, credential harvesting, and data exfiltration. 3

3

This comprehensive automation represents a significant escalation in cyber warfare capabilities, demonstrating how AI can be weaponized to conduct sophisticated attacks at machine speed.

Anthropic characterized this as an escalation of "vibe hacking" that has emerged in recent months, a term that has gained prominence as individuals without coding experience increasingly use generative AI tools to create and deploy code. 1

1

The Dual-Use Dilemma

Anthropic addressed the fundamental question of why it continues developing tools that could enable cyberattacks against the United States. The company argued that the same capabilities allowing Claude to be misused for attacks also make it crucial for cyber defense. 1

1

When sophisticated cyberattacks occur, Anthropic's goal is for Claude to assist cybersecurity professionals in detecting, disrupting, and preparing for future versions of such attacks.

The company's Threat Intelligence team used Claude extensively in analyzing the enormous amounts of data generated during their investigation of this very incident, demonstrating the defensive applications of the technology. 1

1

Broader Implications for Cybersecurity

Chairman Garbarino emphasized the unprecedented nature of this threat, stating that "for the first time, we are seeing a foreign adversary use a commercial AI system to carry out nearly an entire cyber operation with minimal human involvement." 2

2

He warned that this development should concern every federal agency and every sector of critical infrastructure.

Experts warn that the same AI capabilities now powering espionage could accelerate financial theft and cryptocurrency-related crimes. Shaw Walters, founder of AI research lab Eliza Labs, noted that "the terrifying thing about AI is the speed" and that what previously required manual effort can now be automated at massive scale. 3

3