White House Warns of AI Model 'Extraction' Campaigns

Artificial Intelligence & Machine Learning , Government , Industry Specific White House Warns of AI Model 'Extraction' Campaigns Agencies Urged to Track and Disrupt Coordinated AI Extraction Campaigns Chris Riotta (@chrisriotta) • April 24, 2026     Credit Eligible Get Permission Image: Ruben Martinez Barricarte/Shutterstock The White House is planning to ramp up collaboration and information-sharing with U.S. artificial intelligence companies as part of an effort to counter what officials describe as systematic, foreign extraction of sensitive capabilities from domestic AI systems. See Also: Compliance Team Guide for Evasion Prevention & Sanction Exposure Detection White House Office of Science and Technology Policy Director Michael Kratsios said in a Thursday memorandum that the government has evidence foreign adversaries are conducting coordinated campaigns to distill frontier U.S. AI models using proxy accounts and jailbreaking techniques. The memo points to campaigns leveraging "tens of thousands" of distributed accounts to evade detection and rate limits, paired with iterative prompt engineering designed to expose model behavior and underlying system logic. The administration is tasking federal agencies with working with the private sector to develop best practices to identify, mitigate and remediate what it calls "industrial-scale distillation activities." The Trump administration will explore "a range of measures" to hold foreign actors accountable. Campaigns that aim to extract sensitive capabilities from U.S. AI systems do not need to fully replicate frontier models to be effective, officials said. Attackers can approximate performance on targeted tasks or benchmarks, enabling them to commercialize derivative systems while avoiding the cost and time associated with original development. The memo does not spell out specific penalties, but suggests that the White House will begin treating unauthorized model extraction as a form of intellectual property exploitation with national security implications. The administration's push to protect U.S. AI systems from theft and exploitation will likely center in part on expanding telemetry and logging around model interactions, tightening identity and access controls for high-risk users and building detection systems that can flag distributed probing campaigns in real time. The memo also details the role of jailbreak techniques in exposing restricted model outputs and calls for new guardrails that can withstand adversarial prompting without leaking sensitive capabilities or internal alignment signals. Distilled systems can create downstream risk if they lack the same safeguards built into U.S. models, analysts warn, including controls intended to enforce neutrality, reliability and safe use. "AI distillation, when legitimately used to produce smaller, lighter-weight models from more advanced systems, is a vital part" of the AI ecosystem, Kratsios wrote. "Industrial distillation activities that aim to systematically undermine American research and development and access proprietary information, however, are unacceptable."