U.S. Agencies Accuse DeepSeek, Alibaba and Four Chinese AI Firms

The U.S. National Security Agency (NSA), Federal Bureau of Investigation (FBI) and Cybersecurity and Infrastructure Security Agency (CISA) have accused six Chinese artificial intelligence companies, including DeepSeek and Alibaba, of conducting industrial-scale campaigns to extract proprietary capabilities from U.S. AI models.

According to the joint advisory released by the three agencies, DeepSeek, Alibaba, Moonshot AI, MiniMax, StepFun and Z.AI have @coinbureau used knowledge-distillation techniques against U.S. frontier AI systems since at least late 2024. The agencies said the activity was likely conducted with the knowledge of the Chinese government.

DeepSeek and Alibaba Among Companies Named by U.S. Agencies

The advisory said the Chinese companies extracted billions of tokens through millions of exchanges and requests involving U.S. models, including versions of Claude, GPT, Gemini and Grok.

U.S. officials said the activity was designed to obtain proprietary capabilities and reduce the computing, electricity, research and development resources normally required to build frontier AI systems independently. The agencies characterized the campaigns as aggressive and targeted, while acknowledging that knowledge distillation itself is a legitimate technique when used appropriately.

The agencies also said the companies distributed their operations across multiple AI providers, cloud platforms and infrastructure services, making the activity harder to detect. Their advisory recommends that U.S. AI developers strengthen monitoring of suspicious accounts, usage patterns and network activity while sharing intelligence across model providers and infrastructure companies.

Alibaba Previously Accused of Accessing Claude Through Fraudulent Accounts

The latest allegations follow separate claims from Anthropic concerning Alibaba. Bloomberg reported in June that Anthropic accused operators linked to Alibaba’s Qwen AI laboratory of using nearly 25,000 fraudulent accounts to access Claude.

Anthropic said the campaign generated 28.8 million interactions with Claude between April and June and targeted capabilities including software engineering and agentic reasoning.

The accusations add another layer to the growing technology competition between the United States and China, where control over advanced AI capabilities has become closely tied to national security and economic policy.

The immediate issue for U.S. AI companies is how to prevent large-scale extraction of proprietary model capabilities without restricting legitimate markets research and access. The NSA, FBI and CISA have already issued mitigation recommendations, placing greater emphasis on coordinated defenses across the AI infrastructure ecosystem.

Writer: Victoria Hale
Technology & Blockchain Writer

Victoria Hale writes about blockchain technology, digital infrastructure, and the intersection of emerging technologies with finance. Her articles explore how new protocols and systems are shaping the evolving digital economy.

She prioritises clarity and accuracy when explaining technical developments to a general audience.