The House Homeland Security Committee has formally requested that Anthropic CEO Dario Amodei appear before Congress on December 17 to testify about an alleged cyberattack campaign that investigators believe was carried out by China-affiliated threat actors using the company’s AI model, Claude, according to a new report from Axios.
The request was issued by House Homeland Security Chair Andrew Garbarino, a Republican representing New York, who sent letters not only to Amodei but also to Google Cloud CEO Thomas Kurian and Quantum Xchange CEO Eddy Zervigon, asking all three executives to testify as part of the committee’s inquiry. Axios reports that if Amodei agrees to appear, it would mark the first time an Anthropic executive has testified before a congressional committee, a notable milestone for the fast-growing AI company.
The congressional interest follows a Nov. 13 report released by Anthropic, in which the company disclosed that it detected unusual and suspicious activity in mid-September. After conducting an internal investigation, Anthropic concluded that the activity appeared to be part of a “highly sophisticated espionage campaign.” According to the company, the attackers exploited Claude’s agentic capabilities—features that allow the AI to carry out multi-step tasks—to an unprecedented extent, enabling the model to directly participate in executing cyber operations rather than merely assisting human operators.
In its report, Anthropic stated:
“The threat actor—whom we assess with high confidence was a Chinese state-sponsored group—manipulated our Claude Code tool into attempting infiltration into roughly thirty global targets and succeeded in a small number of cases. The operation targeted large technology firms, financial institutions, chemical manufacturing companies, and government agencies. We believe this is the first documented case of a large-scale cyberattack executed without substantial human intervention.”
Anthropic characterized the incident as a significant escalation of what it described as “vibe hacking,” an evolution of the concept of “vibe coding.” Over the past year, “vibe coding” has entered mainstream tech discourse as more non-programmers have used generative AI tools to casually write, modify, and deploy code by describing what they want in natural language rather than writing it manually.
The “vibe” terminology has since expanded into other domains, sometimes controversially. One high-profile example came when Uber co-founder Travis Kalanick appeared on the All-In podcast and claimed he was doing “vibe physics,” suggesting he was making scientific discoveries with the help of AI models—an assertion widely criticized by experts, given that large language models cannot independently generate new scientific knowledge.
The Anthropic report also addressed a question at the heart of the congressional inquiry: why continue developing advanced AI systems if they can be exploited for cyberattacks? The company argued that the same capabilities that enable misuse are also essential for defense. According to Anthropic, Claude plays a crucial role in cybersecurity detection, analysis, and mitigation, and the company claims it has built extensive safeguards into the system.
In its statement, Anthropic wrote:
“This raises an important question: if AI models can be misused for cyberattacks at this scale, why continue to develop and release them? The answer is that the very abilities that allow Claude to be used in these attacks also make it crucial for cyber defense. When sophisticated cyberattacks inevitably occur, our goal is for Claude—into which we’ve built strong safeguards—to assist cybersecurity professionals to detect, disrupt, and prepare for future versions of the attack.”
The company added that its Threat Intelligence team relied heavily on Claude itself to analyze the enormous volume of data generated during the investigation into the alleged espionage campaign.
Chairman Garbarino underscored the seriousness of the situation in a statement to Axios, saying:
“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. That should concern every federal agency and every sector of critical infrastructure.”
When reached by phone on Wednesday, a spokesperson for Anthropic declined to comment on the record regarding the upcoming congressional hearing or Amodei’s potential testimony.