China-aligned TA419 targeted US AI policy experts with session phishing
Dark Reading's Oct. 5 item adds little operational detail, calling TA419 an emerging group that impersonated US officials and built professional relationships with AI-policy experts at US think tanks, universities, and legal organizations, and framing the activity as Chinese cyber espionage without naming individuals, tooling, Japan, defense targets, or confirmed compromises.
Proofpoint says China-aligned TA419 impersonated officials and an Anthropic employee to steal Microsoft 365 sessions from US AI-policy experts.
Proofpoint attributes credential phishing since at least April 2025 to China-aligned TA419 against think tanks, universities, and law firms, with Infosecurity and The Hacker News also naming defense contractors and several outlets including institutions in the United States and Japan; The Record describes only a Chinese actor and does not name TA419, while Dark Reading calls TA419 an emerging group that cultivated professional relationships. From July 8—placed in 2026 by The Register, GBHackers, and Help Net Security—lures impersonated former White House science or OSTP official Lynne Parker, whom The Register and Help Net Security call Lynne Edwards Parker, and economist Heidi Crebo-Rediker; a February campaign dated 2026 by several outlets impersonated an Anthropic employee, described as senior by multiple reports, using a military-use-of-Claude pretext that The Hacker News titles "Request for Feedback on Military Integration of Claude," often after benign outreach some reports tie to a fictitious AI Policy Advisory Committee. Shortened links, sometimes after a Cloudflare Turnstile check, opened spoofed OneDrive pages using Frameless BitB—called modified or customized, Evilginx-style by Cyber Security News, and built on Frameless BitB and Evilginx by The Register—that proxied Microsoft 365 and Entra ID sign-ins and captured passwords, MFA codes, and session cookies. CyberScoop and Cyber Security News say compromises were not confirmed; CyberScoop says domains spoofed the Heritage Foundation, the WEF, and the Japan-Taiwan Exchange Association, and Help Net Security says they were typically registered at NameSilo. Sources disagree on attribution strength: Infosecurity says Proofpoint assesses likely Chinese collection on US AI policy and export controls, GBHackers adds defense integration, CyberScoop says Proofpoint did not directly link the activity to the Chinese government, The Register's headline calls the operators suspected Chinese spies, and Dark Reading frames it as Chinese cyber espionage. Separately, The Record reports Cisco Talos findings that Chinese state-backed operators used the Antino backdoor against governments in Taiwan, India, the Philippines, and five other countries, with about 350 endpoints compromised between September 2025 and July 2026.
- Proofpoint attributes credential phishing since at least April 2025 to China-aligned TA419 against think tanks, universities, and law firms, with some reports also naming defense contractors and targets in the United States and Japan.
- July lures, dated July 8 and placed in 2026 by The Register, GBHackers, and Help Net Security, impersonated Lynne Parker—also reported as Lynne Edwards Parker—and economist Heidi Crebo-Rediker.
- A February campaign dated 2026 by several outlets impersonated an Anthropic employee, called senior by multiple reports, using a military-use-of-Claude lure that The Hacker News titles "Request for Feedback on Military Integration of…
- Shortened links, sometimes after Cloudflare Turnstile, opened spoofed OneDrive pages using Frameless BitB—also described as Evilginx-style or built on both kits—that proxied Microsoft 365 and Entra ID and captured passwords, MFA codes, and…
- CyberScoop and Cyber Security News say compromises were not confirmed; CyberScoop says Proofpoint did not directly link the activity to the Chinese government.
- Reported domains spoofed the Heritage Foundation, the WEF, and the Japan-Taiwan Exchange Association and were typically registered at NameSilo; The Register says Proofpoint advises phishing-resistant passkeys.
- Separately, The Record cites Cisco Talos on the Antino backdoor compromising about 350 government endpoints in Taiwan, India, the Philippines, and five other countries between September 2025 and July 2026.
- Dark Reading calls TA419 an emerging group that built professional relationships with US AI-policy experts and frames the activity as Chinese cyber espionage, without naming tooling or confirmed victims.
Coverage
oldest first · one row per article · UTCChina-Linked Hackers Impersonate AI Experts to Target US Policy Insiders
China-linked TA419 phishes US AI-policy staff using expert impersonation and a Microsoft 365 session-stealing proxy.
AI policy circles targeted in China-linked phishing operation
China-aligned TA419 ran adversary-in-the-middle phishing campaigns impersonating officials and an Anthropic staffer to steal credentials from U.S. AI policy experts.