China-Linked Hackers Posed as AI Experts to Breach US Policy Figures, Report Finds

A newly published cybersecurity report has revealed that a hacking group with suspected ties to China spent months impersonating well-known artificial intelligence figures in an effort to compromise the personal and professional accounts of American policy experts. According to research released Thursday by the cybersecurity firm Proofpoint, the campaign represents a calculated attempt to infiltrate the networks surrounding the people who help shape United States technology and AI policy.

The group, tracked by researchers under the designation TA419, has been active since at least April 2025, according to the report. Rather than relying on technical exploits alone, the operators behind the campaign leaned heavily on deception, building trust with their targets before attempting to harvest their login credentials. This approach, often described as social engineering, exploits human judgment rather than software vulnerabilities, making it particularly difficult for conventional security tools to detect.

At the center of the July phase of the operation was the impersonation of Lynne Edwards Parker, who previously served as principal deputy director of the White House Office of Science and Technology Policy. By adopting the identity of a credible and recognizable figure in the AI policy world, the attackers sought to lend their messages an air of legitimacy that would encourage recipients to respond without suspicion.

Proofpoint's investigators described a multi-stage process designed to lower a target's guard. In the opening stage, victims would receive emails that appeared benign, often containing an invitation to participate in something like an "AI Policy Advisory Committee." These overtures were crafted to appeal to the professional interests of people working in the AI and technology policy space, giving them little reason to be immediately wary.

Once a target engaged with the initial message, the hackers escalated their efforts. The report states that the targets were then directed to a fabricated login page built specifically to capture their usernames and passwords. This credential-harvesting step marks the point at which a seemingly routine professional exchange turns into a genuine security breach.

One particularly sophisticated element of the campaign involved the use of a technique that produces a fake browser pop-up window embedded within a legitimate webpage. Because the fraudulent sign-in prompt appears inside a real site, victims have a harder time recognizing that they are handing their credentials to malicious actors. This method exploits the implicit trust internet users place in the websites they visit regularly.

The scope of the targeting extended well beyond a single country or institution. According to the report, the hackers pursued policy experts based at think tanks, defence contractors, universities, and law firms, with victims located in both the United States and Japan. This broad selection of targets suggests an interest not simply in individual credentials, but in the networks of influence and information that connect AI policy communities across allied nations.

Proofpoint did not publicly identify the specific individuals whose accounts were compromised. However, Reuters was independently able to confirm at least one victim: Alex Engler, a former White House official who now leads the Penn Center on Media, Technology, and Democracy. Engler recounted to Reuters that he received one of the emails but grew suspicious after consulting with colleagues in the industry, ultimately determining that the message had come from an impersonator rather than the person it claimed to be.

The July activity was not the group's first attempt to leverage the credibility of AI sector figures. In February, the same cluster of hackers was linked to the impersonation of a "prominent" employee at Anthropic, the AI company, as part of a continuing effort to reach policy experts. That earlier episode, alongside the more recent campaign, indicates a sustained and deliberate strategy rather than an isolated incident.

Analysts behind the report expect TA419 to keep refining and repeating this playbook, continuing to target think tanks and other policy-focused organizations. Central to that strategy is the ongoing use of real-world experts' identities, which allows the group to exploit name recognition and professional trust to bypass the skepticism that might otherwise greet an unfamiliar sender.

Parker did not respond to a request for comment from Al Jazeera, leaving questions about whether she was aware of the impersonation or had taken any steps in response. The broader silence from many of those targeted is common in such investigations, as victims and their institutions often refrain from public statements while security reviews are underway.

The findings land at a moment of heightened sensitivity around the intersection of artificial intelligence and national security. In recent weeks, attention has focused on a proposed US–China AI "hotline" that President Donald Trump reportedly plans to pitch to Chinese President Xi Jinping, as well as warnings from AI leaders about catastrophic risks and the reluctance of both Washington and Beijing to slow development. The report adds a fresh dimension to those tensions by underscoring how the AI policy community itself has become a target.

For security professionals, the campaign illustrates a recurring lesson: even well-informed experts can be fooled when an attack is tailored to their professional world. The use of fake advisory committees, imitation of respected officials, and deceptive login pages all rely on context and credibility rather than technical trickery, making awareness and verification essential defenses.

As governments and technology companies continue to debate how best to govern artificial intelligence, the Proofpoint findings suggest that the people drafting and advising on those policies are themselves vulnerable. The report's authors believe the group will persist in targeting think tanks and policy specialists, meaning that vigilance within these communities will remain a critical line of protection against future intrusions.