Feds Accuse China of 'Systematic' Distillation of US AI Models
The NSA, CISA, and FBI are accusing (PDF) several Chinese AI companies of carrying out "industrial-scale" distillation campaigns against leading U.S. models such as ChatGPT, Claude, Gemini, and Grok. Since at least 2024, the companies have allegedly routed millions of requests across accounts, APIs, proxies, cloud providers, and third-party aggregators to extract capabilities for their own…
Authorities have accused several Chinese AI companies of systematically extracting functionalities and capabilities from leading U.S. AI models through large-scale knowledge distillation campaigns. The National Security Agency, Cybersecurity and Infrastructure Security Agency, and Federal Bureau of Investigation allege that firms such as DeepSeek, Moonshot AI, and others have been engaging in these activities since at least 2024.
These campaigns involve routing millions of requests through various accounts, APIs, proxies, cloud providers, and third-party aggregators to extract desired features and functionalities from prominent AI models like ChatGPT, Claude, Gemini, and Grok. For instance, DeepSeek utilized four versions of Claude, two versions of Gemini, five versions of ChatGPT, and Grok 4 to train its R1 and R3 models, while Moonshot AI distilled 18 U.S. models, including Anthropic's advanced commercially available model, to train its Kimi-K2 and Kimi K3 models.
To avoid detection, Chinese AI companies employ sophisticated tools and systems to route requests through multiple pathways, using different accounts, models, and platforms, and leveraging proxies and gray tech markets to circumvent geographic restrictions and terms of use.
Written by urgent.news from Slashdot's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
