Chinese state-linked hackers spoof Anthropic staff to target US AI policy experts

Cybercriminals identified as a Chinese government-aligned group have impersonated senior figures at Anthropic and former US government officials in an attempt to extract sensitive information regarding American artificial intelligence development. The operation, detailed in a recent report by cybersecurity firm Proofpoint, targeted policy experts at think tanks, universities, and law firms. The attackers used fabricated professional identities to establish trust before deploying malicious software to compromise cloud accounts.

The hacking group, designated as TA419, crafted convincing introductory emails that invited targets to join a non-existent AI Policy Advisory Committee or contribute to a fictional Senate Committee on Foreign Relations report. After establishing a baseline of normal correspondence, the actors sent documents containing malware designed to gain access to victims’ cloud infrastructure. In one specific instance, the perpetrators impersonated a senior Anthropic employee and used the subject line Request for Feedback on Military Integration of Claude to lure a think tank worker into opening the malicious payload.

Mark Kelly, a researcher at Proofpoint, stated that the group is a Chinese government-aligned threat actor. This assessment is based on targeting patterns that align with Chinese government interests, observed technical infrastructure, and corroboration from industry partners. The report highlights the sophistication of the social engineering tactics used to bypass security protocols. The group specifically focused on individuals with expertise in AI policy and regulation, indicating a strategic interest in understanding the US approach to governing emerging technologies.

Lynne Edwards Parker, who served as Principal Deputy Director of the White House Office of Science and Technology Policy under President Joe Biden, was among those whose identity was spoofed. She expressed concern for colleagues and noted a sense of sadness that professional relationships built over her career were being exploited by bad actors to deceive others. The incident underscores the vulnerability of high-profile figures in the AI sector to targeted espionage campaigns.

The report emerges amid a period of significant debate over AI regulation in the United States. While some industry leaders, including Anthropic’s Dario Amodei, have called for government oversight to prevent potential security and infrastructure catastrophes, the White House and figures such as Meta’s Mark Zuckerberg and Nvidia’s Jensen Huang have resisted such measures. The administration recently hosted a summit where industry leaders signed a document described by President Donald Trump as a morally-binding agreement for self-policing. Trump has argued that the US must maintain its lead in the international AI race against China, stating that whoever wins AI wins. Proofpoint warns that TA419 will likely continue targeting experts in technologies and geographies of interest to the Chinese government, continuing to spoof the identities of real subject-matter experts.