China-Linked Hackers Pose as AI Experts to Breach US Policy Targets, Report Finds

A China-aligned hacking group has been masquerading as prominent artificial intelligence specialists in a sustained effort to compromise the email accounts of American policy experts, according to new research from the cybersecurity firm Proofpoint. The report, released on Thursday, details a campaign that blends social engineering with credential-stealing technology to penetrate the networks of influential figures shaping technology and national security policy.

The group, tracked by Proofpoint as TA419, has been active since at least April 2025, according to the firm's findings. Rather than relying on software vulnerabilities, the hackers have built their strategy around deception, impersonating real people with genuine credentials in the AI field to win the trust of their targets before striking.

According to the report, the operation reached a notable peak in July, when the hackers set their sights on a group of US policy experts while posing as well-known figures in the AI community. Among the identities they adopted was that of Lynne Edwards Parker, the former principal deputy director of the White House Office of Science and Technology Policy—a role that would lend instant credibility to any correspondence about AI policy matters.

The mechanics of the scheme follow a pattern that security researchers describe as increasingly common in state-linked espionage. The hackers begin by sending innocuous-looking emails designed to start a conversation, often with a hook tailored to the professional interests of their targets. In this case, the initial messages contained invitations to join an "AI Policy Advisory Committee," a plausible-sounding initiative for someone working in the field.

Once a recipient engaged with the message, the attackers escalated their approach by directing the target to a counterfeit login page crafted to harvest usernames and passwords. The report highlights a particularly sophisticated technique used in the campaign: a fake browser pop-up window embedded within a legitimate webpage that mimics an authentic sign-in prompt. This method makes it significantly harder for victims to recognize that they are surrendering their credentials to malicious actors.

The scope of the targeting extended beyond the United States. Proofpoint's research indicates that the hackers also pursued policy specialists in Japan, broadening the geographic reach of the operation. The intended victims spanned think tanks, defense contractors, universities, and law firms—sectors where access to sensitive policy discussions and proprietary research could offer valuable intelligence to a foreign government.

While Proofpoint did not identify the specific individuals who were targeted, Reuters was able to confirm at least one of them independently. Alex Engler, a former White House official who now leads the Penn Center on Media, Technology, and Democracy, acknowledged receiving one of the fraudulent emails. He told Reuters that after consulting with colleagues in the industry, he determined that the message had come from an impersonator rather than the person it claimed to be.

The report notes that this is not the first time TA419 has employed such tactics. In February, the same group orchestrated the impersonation of a "prominent" employee of Anthropic, the AI safety company, as part of a broader effort aimed at policy experts. That earlier operation demonstrates a consistent pattern: the hackers repeatedly adopt the identities of real, recognizable figures to lend authenticity to their overtures.

Proofpoint assesses that TA419 will continue to pursue think tanks and other policy-focused organizations, and that the impersonation of genuine experts will remain a central element of its toolkit. The firm's analysts believe the group sees value in leveraging the reputations of credible individuals to bypass the skepticism that might otherwise greet unsolicited communications.

The findings underscore a widening intersection between artificial intelligence policy and national security concerns. As governments in Washington and Tokyo accelerate their work on AI regulation, standards, and strategic competition, the policy experts involved in those discussions become attractive targets for intelligence-gathering operations. Their inboxes can contain drafts of forthcoming regulations, internal deliberations, and contact networks that are difficult to obtain through other means.

The campaign also reflects a broader shift in cyber-espionage tactics. Instead of exploiting technical flaws in software, attackers increasingly rely on human psychology, crafting messages that appear legitimate and urgent enough to prompt a click. The use of a fake pop-up login window inside an otherwise genuine webpage illustrates how attackers are refining their methods to defeat the visual cues that users have been trained to look for.

For organizations and individuals working in AI policy, the report carries an implicit warning to verify the identities of anyone requesting credentials or directing them to a login page, even when the request appears to come from a trusted name in the field. Proofpoint's research suggests that the impersonation of real experts is likely to persist as long as it continues to produce results.

Al Jazeera reached out to Lynne Edwards Parker for comment on the report, but she did not respond to the request. The lack of a public response is not uncommon in cases involving impersonation, where the individuals whose identities are borrowed may themselves be victims of the scheme.

The report arrives amid heightened attention to cybersecurity in the AI sector, with recent headlines covering everything from proposed US-China AI hotlines to regulatory probes of major AI companies. As the policy landscape grows more consequential, the incentives for espionage in this space are likely to grow in tandem, making vigilance a necessity rather than an option for those working at the intersection of technology and government.