China-Linked Espionage Group Targets AI Policy Experts
Proofpoint researchers identified a China-aligned cyber espionage operation targeting U.S. artificial intelligence policy experts through sophisticated phishing campaigns. The activity, attributed to a threat actor tracked as TA419, began in July 2024 and involved impersonating high-profile officials and experts to establish trust before attempting credential theft.
Initial Contact and Trust-Building Tactics
The campaign opened with emails that impersonated Lynne Parker, former principal deputy director of the White House Office of Science and Technology Policy, and Heidi Crebo-Rediker, an economist and foreign policy expert. These messages invited recipients to join a supposed AI policy advisory committee or contribute to a report on AI export controls and semiconductor supply chains. Notably, the initial emails did not contain malicious links or direct requests for passwords, instead focusing on engaging the target in dialogue.
Adversary-in-the-Middle Attack Framework
After a target responded, TA419 sent a shortened URL that redirected through multiple websites before landing on a counterfeit Microsoft OneDrive sign-in page. Proofpoint classified this as an adversary-in-the-middle (AiTM) phishing attack. In such attacks, victims interact with legitimate Microsoft infrastructure during parts of the login process — entering passwords, completing MFA prompts, and passing access checks — while the attacker covertly captures session tokens and authentication data.
Use of Modified Frameless BitB Tool
To enhance deception, TA419 employed a customized version of the open-source Frameless BitB tool. This tool injects a fake browser window within a legitimate webpage, mimicking a trusted sign-in interface. In this case, it presented a fraudulent Microsoft login prompt over a page designed to resemble a OneDrive document-sharing site, increasing the likelihood of credential disclosure without raising suspicion.
Targeting Anthropic Employee in February Campaign
Proofpoint also uncovered a February 2024 campaign where TA419 impersonated a senior employee of Anthropic, the AI company behind the Claude model series. The phishing message was sent to an AI policy analyst at a U.S. think tank, requesting feedback on the military’s use of Claude AI models — a topic of significant debate at the time due to concerns over AI safety and national security implications.
Broader Targeting Patterns and Infrastructure
TA419’s targeting extends beyond AI policy, with Proofpoint noting activity since at least April 2024 against individuals linked to U.S. and Japanese think tanks, defense contractors, universities, and law firms. This aligns with the group’s historical focus on defense, national security, energy, international relations, and foreign policy sectors.
The actors registered domains designed to mimic legitimate organizations, including those resembling the Heritage Foundation, the World Economic Forum, and the Japan-Taiwan Exchange Association, further supporting their social engineering efforts.
Attribution and Geopolitical Context
While Proofpoint attributes the activity to TA419 and notes its alignment with Chinese interests, the report does not assert direct ties to the Chinese government. The findings come amid heightened U.S.-China strategic competition over AI development, export controls, semiconductor supply chains, and military applications of AI. Both nations have exchanged accusations regarding cyber operations, with Washington alleging that Beijing distills U.S. AI models to power open-weight systems operated by Chinese firms — a claim Beijing denies.
Proofpoint has published indicators of compromise associated with this campaign on its website for defenders seeking to detect or block related activity.
Source: CyberScoop