The Hacker News reports that a China-nexus espionage group tracked as TA419 has run multiple credential phishing campaigns against artificial intelligence experts at US think tanks and universities. The lures imitate Microsoft sign-in pages and are built as adversary-in-the-middle proxies.

An adversary-in-the-middle page does not guess a password. It sits between the victim and the real Microsoft login, passes everything through, and keeps the session token that comes back. The victim types the correct password and the correct one-time code. Both are valid, because both went to Microsoft.

The multifactor prompt was supposed to be the control here. In this case it is part of the route.

The targeting is specific. People who draft AI policy for a think tank see export-control thinking, draft positions and the names of the officials they are talking to months before any of it is published. Their mailboxes are the product.

The same week, MI5 said that more than 100 UK-linked academics had taken part in research funded by China's Ministry of State Security, according to the same outlet. A group phishing for the credentials of AI researchers and an intelligence service paying them for research are two collection methods pointed at one population.

The desk has no public evidence in hand tying TA419 to any ministry, and the reporting attributes it only as China-aligned. Attribution at that level is a label for a cluster of activity. It is not a finding about who tasked it.

Any institution with AI researchers has an exposure it can check this morning: which accounts still authenticate with an SMS or authenticator-app code. A FIDO2 security key or passkey is bound to the real sign-in domain, so it will not answer a proxy page. For this campaign, that control would have changed the outcome.