The White House on Thursday accused foreign actors—primarily based in China—of running “industrial-scale” campaigns to copy the capabilities of U.S. AI models, warning the operations threaten American innovation and national security.
In a memorandum titled “Adversarial Distillation of American AI Models,” Michael Kratsios, Assistant to the President for Science and Technology and Director of the Office of Science and Technology Policy (OSTP), said the U.S. government has evidence of coordinated efforts to “distill” frontier American AI systems. Kratsios told followers on X that these campaigns use “tens of thousands of proxy accounts” and jailbreaking techniques to systematically extract proprietary model behavior. “We will be taking action to protect American innovation,” he wrote.
What’s a distillation attack? Operators treat a high-performing model as an oracle: they feed it many prompts, capture its outputs, and use those input-output pairs to train a smaller, cheaper model that mimics the original. While the knockoffs may not fully match the source model’s performance, they can hit comparable benchmark scores at a fraction of the cost—and, according to the administration, can strip away safety controls meant to keep systems neutral and truthful.
U.S. AI firms have already raised alarms about such tactics. In February, Anthropic accused Chinese labs DeepSeek, Moonshot, and MiniMax of harvesting millions of responses from its Claude model using roughly 24,000 fraudulent accounts to train competing products.
The administration said it will coordinate federal agencies and U.S. AI companies to bolster protections around frontier models, help industry develop defenses against large-scale distillation campaigns, and pursue ways to hold foreign actors accountable. The memo also acknowledged lawful distillation’s role in creating smaller open-source models, but drew a clear line: unauthorized, systematic copying of American AI innovations is unacceptable. “There is nothing innovative about systematically extracting and copying the innovations of American industry,” the memo states.
Why this matters to crypto: AI powers trading algorithms, automated on‑chain analysis, smart-contract auditors, bot-driven market activity, and security tools. If high-quality proprietary models are illicitly replicated, those capabilities could be broadly disseminated—amplifying both innovation and the risk of misuse in crypto markets and tooling.
The OSTP did not immediately respond to a request for comment by Decrypt.
Read more AI-generated news on: undefined/news
