WASHINGTON — A China-linked cyber-espionage group impersonated former US government officials and artificial intelligence specialists in a targeted phishing campaign aimed at researchers and policy experts working on AI, according to cybersecurity firm Proofpoint.
The campaign sought to steal login credentials and potentially gain access to emails and cloud accounts belonging to people involved in sensitive areas of US AI policy, including regulation, export controls and national strategy.
Proofpoint identified the group as TA419 and said it had been conducting similar operations against organisations in the United States and Japan since at least 2025. The targets have included think tanks, universities, defence contractors and law firms.
In the latest campaign, hackers posed as Lynne Parker, a former senior official at the White House Office of Science and Technology Policy. Parker previously served as principal deputy director of the office and was involved in US government work on artificial intelligence.
The attackers used Parker's identity to send emails inviting AI policy experts to participate in a purported “AI Policy Advisory Committee”. Other messages proposed contributions to what appeared to be a Senate Foreign Relations Committee report examining AI export controls and supply chains.
The messages were designed to establish credibility before directing recipients towards websites intended to capture their Microsoft login credentials.
Proofpoint said the attackers used a phishing technique that can make a fraudulent login window appear within a legitimate webpage, potentially making it more difficult for victims to recognise that their credentials are being captured.
At least one of the targeted individuals was identified as Alex Engler, a former White House official who now heads the Penn Center on Media, Technology, and Democracy. Engler told Reuters that he received an email appearing to have been sent by Parker inviting him to participate in an AI policy project.
He became suspicious of the message and contacted others in the field before determining that it was fraudulent.
The number of people targeted appears to have been small. Proofpoint said fewer than 10 individuals across a limited number of organisations were identified in the campaign.
The company said the narrow targeting suggested that the operation was focused on gathering intelligence about US AI policymaking rather than simply stealing technology or conducting large-scale credential theft.
The group also impersonated Heidi Crebo-Rediker, a former State Department official and economist, in a separate campaign. In February, it reportedly used the identity of a senior Anthropic employee in an attempt to target an AI policy analyst.
Proofpoint attributed the activity to China based on the group's technical infrastructure, malware and targeting patterns, which the company said were consistent with Chinese intelligence-gathering priorities.
The Chinese Embassy in Washington did not immediately respond to requests for comment. Beijing has repeatedly denied accusations that it conducts state-sponsored cyber espionage.
The campaign comes as Washington and Beijing compete over the development and regulation of advanced artificial intelligence. Both governments regard AI as an increasingly important economic and national-security technology, while disputes over advanced semiconductors, export controls and access to computing infrastructure have intensified.
That competition has created a growing pool of sensitive information around US AI policy. Researchers working on export restrictions and national AI strategies can have access to information that may help governments and companies understand Washington's future regulatory and technology policies.
Parker said the attempted impersonation demonstrated how professional relationships can themselves become targets in cyber operations.
The campaign did not rely primarily on technical vulnerabilities in AI systems. Instead, the attackers exploited trust by assuming the identities of people already known or credible within the AI policy community.
Proofpoint said it expects TA419 to continue targeting policy experts and organisations working on technologies considered strategically important to China.
The group has also used impersonation tactics involving other organisations and officials, indicating that its operations extend beyond the latest campaign against US AI specialists.
The attacks illustrate a broader shift in cyber-espionage towards highly targeted social engineering, where attackers seek access by manipulating individuals rather than directly compromising heavily protected systems.
For US AI researchers and policymakers, the episode adds another layer of security concerns to an already intense technological competition with China. The immediate investigation remains focused on determining whether any targeted accounts were successfully compromised and whether sensitive information was obtained.
