Artificial Intelligence · Cybersecurity
TA419 targets U.S. AI policy experts with phishing
Proofpoint said the China-aligned TA419 group ran credential phishing campaigns against AI policy experts and other researchers, using impersonation, redirect chains and an adversary-in-the-middle page to capture Microsoft logins.
- Proofpoint said TA419 had run several credential phishing campaigns against AI experts at U.S. think tanks, universities and law firms.
- The campaigns impersonated economists, AI policymakers and an Anthropic employee, and one February 2026 attempt targeted an AI policy expert at a U.S. think tank.
- Proofpoint said the phishing route used shortened links, a multi-stage redirect and a Cloudflare Turnstile check before reaching a OneDrive adversary-in-the-middle page.
- Wael Masri had described Frameless BitB in January 2024, and Proofpoint said TA419 had extended it with telemetry and automation that tracked Microsoft sign-ins.
Proofpoint said TA419 had been behind repeated credential phishing activity aimed at AI specialists working for think tanks, universities and law firms. It said the group had operated since at least April 2025 and had also gone after defence contractors.
In one case in February 2026, the attackers posed as economists and AI policy figures, as well as an Anthropic employee, to reach an AI policy expert at a U.S. think tank. Around July 2026, Proofpoint said the group also used the identity of a former White House Office of Science and Technology Policy staff member in similar attempts.
The lure began with an innocuous message intended to win trust. If the target replied, the attacker sent a shortened link that led through a chain of redirects and then to a OneDrive login page after a Cloudflare Turnstile check.
Proofpoint said the page used Frameless BitB, a browser-in-the-browser variant, and that TA419 had added a bespoke module for telemetry and automation. That module tracked the Microsoft sign-in process, passed captured details to the attacker and relayed them to Microsoft infrastructure so the login appeared to succeed.
Named in this story
People
- Wael Masri
- Described Frameless BitB in a January 2024 note
Companies
- Proofpoint
- Published the analysis that attributed the activity
- Anthropic
- Its employee was impersonated in one of the campaigns
- Microsoft
- Its sign-in flow and infrastructure were mimicked
Organisations
- TA419
- A China-aligned group attributed to the phishing campaigns
Governments and agencies
- White House Office of Science and Technology Policy
- A former member of its leadership team was impersonated
Products and systems
- Cloudflare Turnstile
- The phishing chain required its check before redirection continued
- OneDrive
- Hosted the adversary-in-the-middle phishing page
How the source tells it
The source used a measured security-analysis register, with its main emotive pressure coming from framing the phishing as part of China-U.S. espionage competition.
- threat framing loaded attribution to a China-aligned espionage group and wider strategic rivalry cast the activity as part of a larger security threat