GoKawiilUS intelligence agencies name DeepSeek, Alibaba, five others in AI model theft warning
The NSA, CISA, and FBI jointly identified six Chinese AI companies—DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI—as running large-scale campaigns since late 2024 to extract the underlying capabilities of leading US AI systems like Claude, GPT, Gemini, and Grok. The agencies said these firms likely operated with the Chinese government's awareness and used tactics such as mass fake-account API abuse and prompt injection to expose hidden reasoning processes.
GoKawiil's interpretation of the reporting above, not reported fact.
If proven at the scale described, this kind of distillation could let Chinese firms replicate frontier AI performance while skipping the enormous compute and research costs American labs have sunk into training, narrowing the US technological edge. The government's call for coordinated industry-wide defenses signals that Washington now treats model theft as a national security issue on par with traditional IP espionage, likely prompting new API safeguards and account verification across US AI providers.
- Six Chinese AI firms named: DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI.
- Tactics include mass fake accounts hitting APIs and prompts designed to expose hidden chain-of-thought reasoning.
- US agencies urge industry-wide coordination to protect frontier model capabilities from large-scale extraction.
The Coming Wave" by Mustafa Suleyman (hardcover) — If headlines about frontier-model theft and the US-China AI race have you wanting the bigger picture, this book from the DeepMind co-founder digs into how fast AI capabilities spread and why containing them is so hard. It's a readable way to understand the stakes behind distillation, national security worries, and the scramble for AI leadership.
See The Coming Wave" by Mustafa Suleyman (hardcover) on Amazon → Affiliate link — we may earn a commission on purchases, at no extra cost to you. Product picked by AI based on this article; it is not a tested recommendation.Source: arstechnica.com, 2026-09-09
Published there as: “Six Chinese AI firms accused of aggressively copying US frontier models”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.