Hackers Posed as an Anthropic Executive to Target AI Policy Experts
Proofpoint says China-aligned group TA419 impersonated a senior Anthropic employee and a former White House OSTP official in credential-phishing campaigns targeting US AI policy researchers, though no Anthropic system, model or account was actually breached.
The people who shape AI policy, not the models themselves, are now the target. Cybersecurity firm Proofpoint reported that China-aligned group TA419 ran credential-phishing campaigns against AI policy experts at US think tanks, universities and law firms, impersonating both a senior Anthropic employee and former White House OSTP official Lynne Parker to get victims to engage.
It's worth being precise about what actually happened: no Anthropic system, model or account was compromised — an employee's identity was simply borrowed as a convincing lure. The campaign's actual targeting reveals exactly what TA419 was after. A few details from Proofpoint's research stand out:
- In February 2026, the group posed as a senior Anthropic employee and emailed a think-tank policy analyst under the subject line "Request for Feedback on Military Integration of Claude" — a deliberately sensitive topic at the time
- A July 2026 wave impersonated economist Heidi Crebo-Rediker, inviting targets to join a fake AI Policy Advisory Committee or contribute to a report on AI export controls and supply chains
- Victims were routed through a modified open-source tool called Frameless BitB into an adversary-in-the-middle phishing kit mimicking Microsoft OneDrive and Entra ID login pages
Proofpoint described the targeting as narrow but deliberate — fewer than ten individuals across multiple organizations received the messages, reflecting TA419's established interest in policy formulation rather than proprietary model data.
The firm said it expects the group to keep targeting think tanks and policy experts working on technologies of strategic interest to Beijing, framing the activity as part of a broader pattern: a day before this report, separate researchers disclosed a different China-linked campaign using a backdoor called Antino against government organizations across eight countries.
The common thread is pointed — as frontier AI companies and their policy relationships become geopolitically significant, the people advising governments on AI are becoming as valuable a target as the AI systems themselves.
