China-Linked Phish Targets AI Policy Experts

Proofpoint reports a China-aligned espionage group (TA419) targeted U.S. AI policy experts at think tanks, universities, and law firms using phishing emails that impersonated well-known officials and AI industry figures. The attackers tried to start a conversation first, then sent a shortened link that ultimately led to a fake Microsoft OneDrive login page designed to steal cloud credentials and active sessions.

Key findings

• TA419 impersonated “prominent officials, economists and an employee of AI company Anthropic” to target AI policy experts.
• The operation focused on “access to cloud accounts held by people at think tanks, universities and law firms.”
• The attackers used a trust-building workflow: “The initial messages did not immediately request passwords… After a target replied, the group sent a shortened link…”
• Victims were redirected to “a false Microsoft OneDrive sign-in page” intended to capture “account credentials and active browser sessions.”
• Proofpoint described it as “an adversary-in-the-middle phishing attack,” where victims may complete MFA while attackers capture session data.
• Proofpoint says the group used a modified open-source tool: “Frameless BitB.”

Who’s being targeted

• Commonly targeted roles: Think tank staff, AI policy analysts, University researchers and faculty, Legal (attorneys and paralegals), Executives in policy/research organizations.
• Affected industries: Think tanks / policy research, Higher education (universities), Legal services (law firms), Defense contractors.
• Attack channels: email, website.
• Impersonated: Former White House OSTP official / well-known economist & foreign policy expert, Senior Anthropic employee.

Awareness takeaways

• Treat “friendly” outreach emails as suspicious when they try to move you into a link-based workflow after you reply.
• Be cautious with shortened links and multi-redirect chains, these are common ways to hide the true destination.
• Train users that MFA alone doesn’t prevent all phishing, some attacks can steal an active session even after MFA approval.
• When an email claims to be from a prominent official or major company, verify the request through a trusted channel before clicking anything.

Red flags to watch for

• Conversation-first approach designed to build trust before sending a link
• Shortened link that redirects through multiple websites
• Unexpected OneDrive/Microsoft login prompt for a policy collaboration invite
• Highly sensitive/controversial topic used to prompt urgent engagement
• Impersonation of a recognizable AI company employee
• Login request that appears during a feedback request