Chinese Hackers Impersonate Anthropic Employee to Extract AI Secrets

Cybercriminals suspected to be Chinese hackers created an elaborate scheme to steal AI-related information by impersonating Anthropic researchers and former government officials, according to a report by cybersecurity firm Proofpoint. The group, identified as "TA419," targeted AI policy experts at US think tanks, universities, and law firms.

The hackers sent out seemingly innocuous emails inviting recipients to join a non-existent "AI Policy Advisory Committee" or contribute to a phony Senate Committee on Foreign Relations report. After a few exchanges, the imposters sent malware-infected documents to gain access to the targets' cloud accounts. One email impersonating a senior Anthropic employee had the subject line "Request for Feedback on Military Integration of Claude," targeting a think tank employee.

Proofpoint's researcher, Mark Kelly, believes the hacking group is a "Chinese government-aligned threat actor" based on the alignment with China's interests and the nature of the attacks. Former White House Principal Deputy Director of the Office of Science and Technology Policy, Lynne Edwards Parker, expressed concern over the scheme exploiting her career relationships.

The issue of AI regulation is a contentious topic in the US, with some experts and executives advocating for government intervention, while the White House and certain industry figures oppose it. The US administration recently hosted a summit where AI professionals signed a "morally-binding" agreement, described by President Trump as a way to maintain the country's lead in the AI arms race against China.

Proofpoint warns that TA419 will likely continue its targeting campaigns, spoofing the identities of real subject-matter experts.