Proofpoint says a China-aligned, espionage-motivated actor it tracks as TA419 ran credential phishing campaigns in July 2026 that impersonated prominent people in US AI policy. In its research post, Proofpoint describes emails posing as a former senior White House Office of Science and Technology Policy official and as a prominent economist, sent to AI policy experts at US think tanks, universities and law firms. It adds that the group impersonated an Anthropic employee in February 2026 to target an AI policy analyst at a US think tank.
The first message was harmless. It invited the target to join a fictitious “AI Policy Advisory Committee” or to contribute to a Senate Committee on Foreign Relations report on AI export controls and supply chains. A target who replied got a shortened link that led through a redirect chain to a fake OneDrive page built to steal Microsoft 365 credentials. Proofpoint says the page relays the real Microsoft sign-in, so the password, the MFA code and conditional access checks all succeed while the attacker captures the session. It says the group has run regular campaigns against think tanks, defense contractors, universities and law firms in the US and Japan since at least April 2025, and that the activity had not been reported publicly before.
Why it matters: the lure works because it matches the target’s job, so awareness training that flags odd requests will not catch it. Proofpoint’s advice is phishing-resistant, origin-bound authentication such as passkeys, and independent verification of unexpected outreach through another channel. Our read: when sign-in is relayed in real time, a one-time code gives no protection, so an incident response plan for a suspected hit should revoke active sessions and tokens, not only reset the password. This is the same reply-first approach we saw in Microsoft’s report on Star Blizzard, and our piece on Britain’s passkey rollout covers the control Proofpoint recommends.
