China-Linked Hackers Impersonate AI Experts to Target US Policy Insiders

* A China-aligned group posed as AI policy figures and economists to steal login credentials from specialists at US think tanks, universities, and law firms.
* Proofpoint tracked the group as TA419.
* Credential phishing activities began in at least April 2025.
* Attackers sent emails impersonating Lynne Parker (White House Office of Science and Technology Policy) and Heidi Crebo-Rediker.
* An earlier incident involved the use of an identity posing as a senior Anthropic employee to contact a think-tank analyst.
* Initial phishing attempts invited responses to join a made-up "AI Policy Advisory Committee" or assist with a Senate Committee on Foreign Relations report.
* Responses led victims to spoofed OneDrive login pages via redirects.
* The attack involved an adversary-in-the-middle (AitM) reverse proxy that captured Microsoft 365 session cookies during login.
* The attackers monitored the login process, automatically accepted "Keep me signed in" settings, and captured one-time codes upon acceptance.
* Proofpoint believes the campaigns feed Chinese intelligence gathering on US AI policy development regarding export controls and model distillation.
* Organizations were advised to adopt phishing-resistant sign-in methods like passkeys.

The pattern emerging here is the weaponization of perceived expertise as a means of access, specifically targeting knowledge domains highly relevant to geopolitical competition, such as AI policy. The group’s focus on impersonating specific roles within government and academia suggests a sophisticated long-term strategy where high-value targets are sought through deception rather than broad, indiscriminate attacks. This mirrors historical patterns where state actors leverage information asymmetry—the gap between what an organization knows and what an external adversary believes it knows—to gain strategic advantage. The linkage to defense, foreign policy, and energy suggests that the operational focus is not just data exfiltration but influence gathering concerning strategic technological trajectories.
The implication for human agency rests on the necessity of treating unsolicited subject-matter outreach as a potential pretext, shifting the baseline expectation from assuming benign communication to demanding independent verification channels before engaging with sensitive information. The failure point identified by this threat is the reliance on social context and institutional authority for initial trust. If adversaries can seamlessly impersonate high-ranking policy experts across multiple domains (policy, economics, technology), it creates an environment where organizational security protocols—which often focus on known technical threats—must expand to incorporate the psychological and contextual vulnerability of specialized human knowledge. The counter-play requires moving beyond technological controls to establish robust epistemic resilience, demanding that individuals recognize the necessary pause required when context suggests a high-stakes negotiation is underway.
Bridge Questions:
What institutional mechanisms are most effective at verifying the legitimacy of unexpected communications that touch upon sensitive policy domains? How can organizations build trust in independent verification channels that bypass standard communication protocols? If expertise impersonation becomes the norm, what new forms of contextual validation are required for accessing high-level strategic knowledge?