|
Getting your Trinity Audio player ready...
|
Chinese Hackers Posed as Former US Officials and an Anthropic Employee to Target AI Policy Experts
A Chinese-linked hacking group has spent months impersonating well-known figures in American AI policy circles, including two former government officials and a senior employee at Anthropic, in an effort to steal email credentials from researchers working on AI regulation, export controls and national strategy, according to a report published Thursday by cybersecurity firm Proofpoint.
Proofpoint, which tracks the group under the designation TA419, said it has watched the hackers try to steal passwords from people at US and Japanese think tanks, defense contractors, universities and law firms going back as far as 2025. The firm attributed the campaign to Chinese state-linked actors based on the specific malware used, the internet infrastructure behind the attacks, and a pattern of targets that lines up closely with known Chinese intelligence collection priorities. Proofpoint didn’t name most of the individual targets, describing them generally as experts working on AI regulation, export controls and national AI strategy.
Reuters independently identified one of the people targeted: Alex Engler, a former White House official who now leads the Penn Center on Media, Technology, and Democracy. Engler said an email he received felt slightly, nebulously off, and after checking with colleagues in the field, he realized he was dealing with an impostor rather than a genuine contact.
The impersonation scheme relied on borrowing the credibility of real, recognizable names in the AI policy world. According to Proofpoint, the hackers posed as Lynne Parker, who held senior technology roles in both the Trump and Biden administrations and now serves as associate vice chancellor emerita at the University of Tennessee, Knoxville. They also impersonated a former State Department economist and, separately, a senior employee at Anthropic, the American AI company whose safety-focused approach has at times put it at odds with the current administration’s looser stance on AI oversight. Beginning around July 8, the hackers sent messages under Parker’s name, and later under the economist’s name, inviting targets to join a fictitious AI Policy Advisory Committee or to contribute to a fabricated report supposedly tied to the Senate Foreign Relations Committee covering AI export controls and supply chains. In a separate incident back on February 26, hackers posing as the Anthropic employee emailed an AI policy analyst at a US think tank with the subject line “Request for Feedback on Military Integration of Claude,” a reference to Anthropic’s AI model, and used the exchange to try to steal the analyst’s email credentials. In at least one case, a fake version of Parker followed up an initial message with a malware-laced document offering supposedly more detail on the invented advisory committee.
Parker learned about the scheme the way many impersonation victims do, through other people. She said colleagues began reaching out after receiving suspicious emails claiming to be from her, and that her first instinct was concern for whoever might be targeted next, even though there was no practical way to warn people she didn’t know were being approached. She described feeling a genuine sense of loss at watching professional relationships she’d built over years get used as a tool to deceive other people, and argued that protecting the people and institutions whose expertise shapes AI policy decisions is itself a matter of national security and economic competitiveness. Engler was reportedly one of at least two people who received suspicious messages impersonating Parker in early July alone, giving some indication of how broadly the campaign was cast.
Asked about the suspected Chinese origin of the operation, Parker said the explanation tracked with the broader dynamic already shaping the field, noting that the US and China are in direct competition over AI, and that trying to get people in AI policy circles to reveal their plans, if that was indeed the goal, wasn’t a particularly surprising move for a state actor to attempt.
Beijing has consistently denied US allegations of state-directed hacking, and nothing in Proofpoint’s report amounts to a criminal charge or formal government attribution naming a specific Chinese agency. Still, the campaign fits a pattern cybersecurity researchers have flagged repeatedly this year: Chinese-linked operations reaching into nearly every layer of the American AI ecosystem, from the semiconductor supply chains that power the technology to the individual researchers and policymakers who help shape how it gets regulated and deployed. Analysts who track these campaigns argue that how well the US protects that human layer, not just its chips and data centers, will matter directly to its broader competitive position in AI.
The episode also arrives against a backdrop of unresolved debate inside Washington over how tightly AI development should be regulated. President Trump has resisted calls for stronger guardrails on advanced AI systems, citing concern that overly cautious policy could let China close the gap in the broader AI race, a stance that puts added weight on episodes like this one, where the very experts shaping that policy debate are themselves being targeted for the information they hold.
This isn’t an isolated data point either. A separate breach disclosed earlier this year saw Chinese hackers steal roughly 60,000 emails from US State Department officials focused on East Asia and Pacific affairs, exploiting a stolen access token tied to a Microsoft engineer just ahead of a planned diplomatic trip to China. Taken together with Thursday’s Proofpoint findings, the pattern suggests Chinese intelligence collection around US AI and foreign policy circles has been sustained and wide-ranging, rather than a single opportunistic incident, with the people who advise on these issues increasingly finding themselves as much a target as the technology and policy decisions they’re trying to shape.
