China-aligned TA419 targets AI policy experts with fake Microsoft sign-in windows

What happened

Proofpoint has attributed several credential phishing campaigns to TA419, a China-aligned espionage group. The targets are artificial intelligence experts at US think tanks, universities and law firms. Since July, the group has impersonated well-known figures, including a former White House technology policy official, inviting recipients to join an AI policy advisory committee or contribute to a report on AI export controls.

People who replied were steered to a page styled like a OneDrive document share. A fake browser window drawn inside the page, built with a modified open-source tool called Frameless BitB, showed a counterfeit Microsoft sign-in that captures the live login session.

Who is affected

The campaigns were narrow: fewer than 10 people at a handful of organisations. Proofpoint has tracked TA419 targeting staff at US and Japanese think tanks, defence contractors, universities and law firms since at least April 2025.

What to do now

• Brief people who work on AI policy, research or export controls about collaboration invitations from unfamiliar addresses.
• Prefer phishing-resistant sign-in methods, such as passkeys or FIDO2 security keys, which adversary-in-the-middle pages cannot replay.
• Watch identity logs for sign-ins from new locations straight after a user opens a shared-document link.

Why it matters for SOC teams

Adversary-in-the-middle phishing steals a working session, so standard multi-factor codes do not stop it. The useful signals are in identity and email logs, read together: an unusual invitation, a sign-in from new infrastructure and a new session token, all within minutes.