The US Treasury Secretary has threatened sanctions against Chinese AI models over allegations of intellectual property theft, according to TechCrunch AI, marking a significant escalation in the regulatory dimension of US-China artificial intelligence competition.
The warning specifically targets Chinese open-source AI models, representing the first time US financial sanctions have been explicitly threatened against AI model developers rather than chip manufacturers or cloud infrastructure providers. The move signals a fundamental shift in how Washington intends to weaponise trade policy in the AI sector.
The Treasury’s position centres on claims that Chinese AI developers have incorporated proprietary training data, architectures, or methodologies from US companies without authorisation. Whilst specific models were not named in the initial threat, the focus on open-source releases suggests concern about freely available Chinese models that compete directly with commercial American offerings.
This approach differs markedly from previous US restrictions, which concentrated on limiting China’s access to advanced semiconductors through export controls on companies including NVIDIA. Those measures, implemented progressively since 2022, restricted sales of high-performance GPUs including the A100 and H100 series to Chinese entities.
The business implications are substantial and multifaceted. US AI companies developing proprietary models—including OpenAI, Anthropic, and Google DeepMind—stand to benefit from reduced competition if Chinese alternatives face market access restrictions. However, the threat also introduces uncertainty for American firms with Chinese operations or partnerships.
Chinese AI developers, particularly those behind open-source models such as DeepSeek and the various Qwen iterations from Alibaba, face potential exclusion from dollar-denominated transactions and US cloud infrastructure. This could fragment the global AI development ecosystem, forcing a bifurcation between Western and Chinese AI stacks.
Enterprise buyers represent another affected constituency. Organisations that have adopted Chinese open-source models for cost efficiency may need contingency plans if sanctions materialise. The threat alone may prompt risk-averse enterprises to avoid Chinese AI technologies regardless of technical merit.
The sanctions threat raises complex enforcement questions. Unlike semiconductor exports, which involve physical supply chains, AI models are digital artefacts that can be distributed through multiple channels. Proving IP theft in AI training—where models learn from vast datasets that may include scraped internet content—presents evidentiary challenges distinct from traditional copyright cases.
Open-source advocates have expressed concern that the precedent could chill collaborative AI development. If governments begin sanctioning models based on contested IP claims, the transparency that characterises open-source AI could become a liability rather than an asset.
The timing coincides with intensifying competition in foundation models. Chinese developers have released increasingly capable open-source alternatives that challenge the performance of proprietary Western models, whilst requiring substantially less computational resources—a development that has unsettled some US AI executives.
Market observers should monitor several developments: whether Treasury provides specific evidence of IP theft; which models, if any, are formally designated for sanctions; how Chinese developers respond, potentially through increased domestic focus; and whether other Western nations adopt similar positions.
The European Union, which has emphasised AI regulation through its AI Act rather than trade restrictions, has not indicated support for the US approach. Divergent transatlantic positions could create regulatory arbitrage opportunities.
This escalation transforms AI development from primarily a technological and commercial competition into an explicitly geopolitical contest, where model releases carry potential sanctions risk. The Treasury’s threat establishes intellectual property enforcement as a new front in the US-China AI rivalry, with implications extending well beyond the immediate parties to reshape how global AI development proceeds.
