Competition in the world of artificial intelligence is starting to look like a technological cold war. On March 13, 2025, OpenAI, the creator of ChatGPT, submitted a bombshell proposal to the US government: ban AI models backed by China, specifically targeting DeepSeek.
This move, part of a climate of US-China rivalry, could disrupt the global AI ecosystem.
A threat named DeepSeek-R1
DeepSeek-R1, the Chinese model that has shaken the tech community in recent months, is at the heart of the controversy. Developed by the DeepSeek lab, this system impresses with its power and low cost, challenging giants like ChatGPT.
But for OpenAI, this feat hides a shadow: the lab is allegedly funded and managed by the Chinese government as part of an ambitious plan to make China the world leader in AI by 2030.
According to OpenAI, this proximity to Beijing poses a major problem: privacy.
Chinese law actually requires companies to hand over sensitive data to the regime if requested. This risk already led Italy to block DeepSeek in its territory shortly after its launch.
However, there is a nuance: while DeepSeek’s online chatbot applies strict censorship, its open-source R1 model remains free to access, allowing anyone to adapt it or run it offline.
A proposal with strategic timing
This initiative from OpenAI is no coincidence. It is part of the “AI Action Plan,” a program launched by the Trump administration to secure the United States’ technological advantage.
In its document submitted to the Executive Office of the President, OpenAI paints an alarming picture: China is investing heavily in AI, closing the gap with American leaders at a worrying speed.
Read more on this topic: Is Chinese AI DeepSeek censored? Here is our test
For Sam Altman’s company, there is an urgent need to act
The parallel with Huawei, the Chinese telecom giant sanctioned by Washington, is striking. OpenAI sees DeepSeek as a strategic tool for Beijing, designed to compete with Western AIs while potentially collecting sensitive data via its API.
But the proposal remains vague: does it target only DeepSeek’s API or also R1, whose open code is already used by platforms like Microsoft Azure or Perplexity? This ambiguity is fueling the debate.
ChatGPT and DeepSeek: Behind the rivalry
This isn’t the first time OpenAI has gone after DeepSeek.
In January 2025, the company accused the Chinese lab of “hacking” its own models, specifically by using ChatGPT to train R1 at a lower cost—an alleged violation of the terms of use. This controversy caused US tech markets to tumble at the time, as investors feared an erosion of OpenAI’s dominance in the face of such an agile competitor.
Today, this legislative offensive seems to confirm a rivalry that is as much economic as it is ideological. Thanks to its performance and accessibility, DeepSeek-R1 is eating up market share, forcing OpenAI to spend a fortune to keep ChatGPT afloat.
Add to that Project Stargate—a $500 billion US investment in AI—and you get a clear strategy: protect national champions while slowing down outsiders.
Regulate, or slow down a rival? OpenAI’s gamble
Far from being limited to a ban, OpenAI is calling for a broader policy. It is asking the United States to promote “democratic” AI, capable of establishing itself globally against models perceived as authoritarian.
To achieve this, the company proposes defending “fair use” in copyright law, a rule that would allow American AIs to train on protected content without legal constraints—a practice that OpenAI says DeepSeek has already used to its advantage.
But this plan raises practical questions
Banning an open-source model like R1, whose code circulates freely, seems unrealistic. And what about international users, where DeepSeek is gaining ground? Some see it as a desperate attempt by OpenAI to slow down a rival that innovates faster and at a lower cost.
See also: Manus: A new Chinese AI in the tech race
What is the future for global AI?
If this proposal succeeds, it could set a precedent in AI regulation, turning a universal technology into a geopolitical battlefield. The United States, along with its allies, could erect digital barriers against China, at the risk of fragmenting the global AI ecosystem.
Meanwhile, DeepSeek continues to thrive outside of US borders, driven by a developer community attracted to its open model.
What do you think of this escalation? Should AI remain a space for collaboration or become a national strategic weapon? We’d love to hear your thoughts in the comments!
