TA419 Targets U.S. AI Policy Experts With Microsoft AitM Phishing

Security firm Proofpoint has profiled a new China-linked cyber espionage group it tracks as TA419. According to the researchers, the actors ran a series of credential phishing campaigns against artificial intelligence experts working for U.S. think tanks, universities and legal organizations.

Who they pretended to be

To lower their targets' guard, the attackers posed as well-known economists and AI policymakers. In February 2026 they impersonated a prominent Anthropic employee to reach an AI policy expert at a U.S. think tank. The lure email carried a telling subject line: "Request for Feedback on Military Integration of Claude."

Around July 2026 the group operated under the names of several individuals, including a former member of the White House Office of Science and Technology Policy leadership team.

How the attack works

Proofpoint assesses that the campaign serves Chinese intelligence interests: Beijing wants to understand how U.S. AI policy and regulation are evolving amid intense strategic competition, mutual accusations of model distillation, and export controls. The group has been active since at least April 2025, previously targeting think tanks, defense contractors and law firms in the U.S. and Japan.

The attack opens with a harmless invitation meant to build trust. Only after the target replies does the next stage fire: a shortened URL launches a multi-stage redirection chain that, after a Cloudflare Turnstile check, lands on a fake OneDrive sign-in page operating as an adversary-in-the-middle (AitM) relay.

Frameless BitB

The page relies on a technique called Frameless BitB, a variant of the browser-in-the-browser attack. The classic version loads a fake login form inside an iframe; the frameless version drops that element and fakes a trusted window using only HTML, CSS and JavaScript. TA419 extended the open-source tool with a bespoke telemetry and automation module that tracks the Microsoft sign-in flow, captures credentials through the AitM proxy, and quietly relays them to the real Microsoft infrastructure.

The method's chief advantage is that the victim notices nothing: the sign-in succeeds, and the stolen session cookies are harvested without any visible sign.

How to defend

Proofpoint recommends that organizations enable phishing-resistant authentication such as passkeys, and that likely targets treat unsolicited subject-matter outreach with caution and verify a sender's identity first. The researchers note that the focus on AI policy experts is an extension of the group's usual remit rather than a departure: TA419 has long gravitated toward defense, national security, energy and international relations.

Source: The Hacker News