Trump Weights Sanctions as China's DeepSeek Closes AI Gap
- US Treasury probes Chinese AI for IP theft
- DeepSeek R1 rivals OpenAI despite chip bans
- Trump administration drafts FDA-style AI vetting
- China plans restrictions on foreign AI access
- Autonomous swarms shift military balance
The United States government is preparing a significant escalation in the ongoing technological war with China, targeting the artificial intelligence sector with potential sanctions.
Treasury Secretary Scott Bessent announced on Wednesday that Washington is investigating whether Chinese AI models have been built using intellectual property stolen from American firms.
Officials said the administration has found what they describe as clear traces of American-made AI models embedded within Chinese counterparts.
Bessent did not mince words during a Senate Finance Committee hearing, stating that the presence of American technology in Chinese models is unacceptable.
This move marks a sharp pivot in how the US intends to police the development of generative AI, moving from export controls on hardware to potential restrictions on software and algorithms themselves.
The investigation centres on the suspicion that Chinese labs are using a technique known as distillation to copy the capabilities of leading US models like GPT-4 and Claude.
- US Treasury Secretary Scott Bessent announced the probe on Wednesday.
- The administration alleges traces of American models in Chinese AI.
- Potential sanctions could target Chinese AI labs and researchers.
The investigation comes as a wave of new releases from Chinese companies has begun to erode the perception of American dominance in the field.
For months, analysts believed that US sanctions on semiconductor exports would cripple China's ability to train advanced models.
However, recent breakthroughs suggest that Chinese engineers have found ways to bypass these hardware bottlenecks, achieving performance levels that rival top-tier American products.
This technological parity has triggered alarm bells within the Trump administration, prompting a review of national security implications.
The White House is particularly concerned that advanced Chinese AI could be used for cyber warfare, surveillance, or to enhance military capabilities that threaten US interests and those of its allies.
Bessent's warning signals that the US is ready to move beyond trade tariffs and into the realm of digital blockades.
DeepSeek R1 and the Distillation Dilemma
At the heart of the dispute is a specific technical process called distillation, which allows a smaller AI model to learn from a larger, more established one.
In January 2025, a Chinese company called DeepSeek released its R1 model, which shocked the industry by matching the performance of OpenAI's best models on common tasks.
What made DeepSeek R1 particularly troubling for Washington was that it achieved this feat without access to the latest high-performance Nvidia chips that are currently banned from export to China.
Industry experts explained that distillation involves using a powerful model to generate vast amounts of training data, which a smaller model then studies to mimic its reasoning patterns.
While this is a standard technique in AI research, US officials argue that Chinese firms are using it to effectively clone American intellectual property without paying for the immense computational cost of discovering those capabilities independently.
- DeepSeek R1 matched OpenAI performance without using banned chips.
- Distillation allows smaller models to mimic larger ones.
- US officials view the practice as IP theft rather than standard research.
The accusation raises difficult questions about the openness of the AI ecosystem.
For years, American researchers have published papers detailing the inner workings of these models, fostering a global collaborative environment.
Now, that openness is being viewed as a vulnerability that strategic competitors are exploiting.
Sources confirmed that the Treasury Department is looking at specific technical signatures—unique quirks in how a model responds to prompts—that suggest a direct lineage to American foundation models.
If these allegations are proven, the US could impose severe restrictions on any Chinese company or individual involved in the development of these models.
This could include cutting off access to American cloud computing services, which remain a vital resource for AI developers globally.
The European tech sector is watching these developments closely, as many European firms rely on the same open-source ecosystems that US officials now want to lock down.
Trump's FDA-Style Plan for AI Vetting
In a striking reversal of his previous deregulatory stance, President Donald Trump is backing an executive order that would subject powerful AI models to government vetting before their release.
National Economic Council Director Kevin Hassett confirmed that the White House is actively drafting this order, which could be signed within weeks.
The proposed framework has been compared to the approval process used by the Food and Drug Administration (FDA) for new drugs.
Under this plan, companies like OpenAI, Google, and Anthropic would need to receive federal clearance before launching new models that exceed a certain capability threshold.
- The White House is drafting an FDA-style AI approval process.
- NEC Director Kevin Hassett confirmed the plan could be signed in weeks.
- The move reverses the administration's earlier push against AI regulation.
This regulatory shift reflects the growing anxiety within the government about the unintended consequences of rapidly advancing AI technology.
While the first term of the Trump administration was characterised by a reluctance to impose rules on the tech sector, the strategic threat posed by China has changed the calculus.
Hassett told reporters that the safety and security of the American homeland necessitate a hands-on approach to AI development.
The order would likely mandate rigorous safety testing to ensure that models cannot be used to create bioweapons, launch cyberattacks, or generate disinformation at scale.
However, critics within the tech industry argue that such a regime could stifle innovation and hand the advantage to Chinese competitors who are not bound by similar restrictions.
They warn that a cumbersome approval process could slow down the release of critical updates, leaving the US vulnerable in a fast-moving race.
Despite these concerns, officials said the administration is determined to proceed, viewing the control of AI as a matter of national sovereignty.
Nvidia Chips and the Beijing Workaround
The effectiveness of US sanctions has been thrown into doubt by the rapid progress of Chinese AI labs.
Since 2022, Washington has placed strict limits on the export of high-end semiconductors, specifically targeting Nvidia's H100 and B200 chips, which are the industry standard for training AI models.
However, recent evidence suggests that Chinese companies have adapted by using older, less powerful chips in massive quantities, or by optimising their code to run more efficiently.
DeepSeek's R1 model, for instance, reportedly relied on a cluster of older Nvidia A100 chips, which were acquired before the tightest restrictions came into effect.
- US sanctions banned the export of Nvidia H100 and B200 chips to China.
- Chinese labs are using older chips or optimising code to compensate.
- China plans to restrict foreign access to its own high-performance AI tech.
Meanwhile, Beijing is preparing its own countermeasures.
Reports from Asia indicate that the Chinese government is drafting regulations that would restrict foreign access to its domestic high-performance AI and semiconductor technology.
This reciprocal action threatens to fragment the global tech market into two distinct spheres of influence.
For European nations, which host major operations of both American and Chinese tech firms, this bifurcation presents a diplomatic and economic nightmare.
Companies like Nvidia, AMD, and Intel have seen their stock prices fluctuate wildly as investors try to gauge the long-term impact of a potential decoupling of the US and Chinese tech sectors.
Analysts noted that while the US currently holds a commanding lead in chip design, China is rapidly catching up in the application layer, finding ways to do more with less computing power.
This efficiency gap is the new front line of the technology race, and it is one where the US is no longer assured of victory.
Autonomous Swarms and the New Battlefield
The race for AI supremacy is not just about economics or consumer chatbots; it is fundamentally reshaping the future of warfare.
Defence analysts are particularly concerned about the application of AI in autonomous swarms—large groups of low-cost drones or robots that can coordinate attacks without human intervention.
According to military experts, AI allows hundreds or even thousands of these autonomous systems to communicate with one another, overwhelming traditional air defence systems through sheer volume.
This technology effectively replaces expensive infantry weapons and heavy artillery with cheap, smart, and disposable units.
- AI enables swarms of autonomous drones to overwhelm defences.
- The technology shifts the economic advantage to the attacker.
- Traditional air defences struggle against volume-based swarm tactics.
The emergence of these systems magnifies the challenge of maintaining a military edge.
If a nation possesses superior AI algorithms, it can field autonomous forces that are faster, more accurate, and more adaptable than those controlled by humans.
Sources within the Pentagon have expressed concern that China's rapid advances in AI could be leveraged to develop these swarm capabilities at a speed that US military procurement cycles cannot match.
Traditional air defence systems are designed to intercept a limited number of incoming threats.
When faced with a swarm of 500 drones acting in concert, these systems can be saturated and rendered useless.
This reality has pushed the technology race to the top of the national security agenda.
The investigation into Chinese AI model theft is therefore not just a trade issue, but a matter of battlefield survival.
Officials believe that if China is allowed to shortcut its development by stealing American code, it could field lethal autonomous weapons years earlier than anticipated.
Europe Caught in the Crossfire of Tech War
As the United States and China dig in for a protracted conflict over artificial intelligence, Europe faces difficult choices about its own digital future.
The European Union has recently passed the AI Act, the world's first comprehensive set of laws to regulate the technology.
However, the aggressive moves by Washington and Beijing may render the European framework less relevant.
If the US imposes an FDA-style approval system and China bans foreign technology,