China-linked hackers impersonated Anthropic staff and ex-US officials to target AI experts

By Splitfeed EditorialLast updated Oct 1, 2026, 20:20 UTC

On October 1, 2026, reporting from Nextgov, CNN, Reuters, and other outlets described a phishing campaign linked to China-based hackers who impersonated a senior Anthropic employee and former White House officials to target US artificial intelligence policy experts. The operation involved spoofed identities of real, named figures inside and close to the US AI policy community, with the goal of stealing login credentials and emails. The specific individuals targeted have not been publicly identified, and no confirmation has been given about whether any data was successfully exfiltrated. China has denied involvement, consistent with its position in previous, similar cyberespionage cases.

The campaign matters because it illustrates how the US-China competition over AI has extended into intelligence operations targeting the human layer of the policy ecosystem, not just the technology itself. The people being phished were not engineers building frontier models but specialists working on questions like export controls and military applications, the exact areas where policy decisions shape the competitive balance between the two countries. That focus suggests the operation was designed to inform Beijing about the US regulatory and strategic posture on AI, rather than to steal model weights or code directly. The operation also highlights the risks created by a small, interconnected expert community where impersonating one trusted identity can open many doors.

On the anti-AI side, the reactions available point in somewhat different directions. Sean Lyngaas, whose account reported the story, noted the targeting was oriented toward military applications of AI and export controls. Five O'Clock Tech framed it as evidence of a clear, strategic intelligence goal around AI. Ross Nesbitt argued the coverage underemphasizes what he described as a systemic vulnerability created by the very specialization these targets hold, suggesting the concentration of expertise itself is a structural risk. Clio2 linked the story to broader concerns about AI companies racing toward superintelligence, citing a Guardian piece quoting a critic who described OpenAI and Anthropic as gambling with lives. Pro-AI voices have not yet reacted publicly; that camp would typically argue that incidents like this demonstrate why the US must accelerate its own AI development and strengthen defenses rather than slow the industry down. The middle ground camp would ordinarily focus on the need for better operational security and interagency coordination without drawing sweeping conclusions about the pace of AI development.

The details most worth watching are whether any of the targeted individuals confirm a compromise, whether the US government attributes the campaign formally to a specific Chinese state-linked group, and whether the episode prompts new security guidance for people working at the intersection of AI and national security policy.