OpenAI is intensifying its campaign to limit the global reach of Chinese artificial intelligence, calling on the U.S. government to ban AI models developed by China’s DeepSeek.
Labeling the company as “state-subsidized” and “state-controlled,” OpenAI warns that DeepSeek’s models pose national security risks, particularly concerning data privacy and potential foreign influence.
In a policy submission to the Office of Science and Technology Policy, OpenAI emphasized the risks of DeepSeek’s AI systems being manipulated by Chinese authorities. The proposal recommends prohibiting such models in sensitive sectors, including government, defense, and critical infrastructure services.
Raising the Alarm: Why OpenAI Sees DeepSeek as a Risk
Central to OpenAI’s concerns is the possibility that DeepSeek’s models could be leveraged to gather sensitive data, manipulate outputs, or even engage in cyber operations aligned with state interests.
According to OpenAI, these systems could be compelled to comply with the demands of the Chinese Communist Party (CCP), risking misuse in activities such as identity theft and the unauthorized extraction of intellectual property.
Chris Lehane, OpenAI’s Vice President of Global Affairs, underlined the stakes by stating that “there is significant risk in building on top of DeepSeek models in critical infrastructure and other high-risk use cases given the potential that DeepSeek could be compelled by the CCP to manipulate its models to cause harm.”
The proposal further suggests that unchecked proliferation of models like DeepSeek’s could compromise democratic processes and threaten global cybersecurity. OpenAI’s concern is rooted in fears that authoritarian regimes may exploit advanced AI to undermine open societies.
A Race Against Time: DeepSeek’s Fast-Tracked Growth
DeepSeek, meanwhile, is racing to strengthen its competitive edge. The company has accelerated the release of its R2 model, initially scheduled for May 2025, and is now expected to debut within weeks. This fast-track decision reflects both intensifying competition and mounting regulatory pressures.
The move also coincides with DeepSeek’s broader strategic response to global scrutiny. The company faces mounting restrictions from Western governments. The U.S. Navy banned the use of DeepSeek AI models on military systems in January, and the state of Texas followed with a similar ban in February, citing security risks.
The Trump administration is expected to prohibit the use of DeepSeek on U.S. government devices due to national security concerns
Italy has also initiated an investigation into DeepSeek’s data privacy practices, probing compliance with GDPR regulations. Compounding these challenges, OpenAI and Microsoft have reportedly launched internal reviews to investigate whether DeepSeek had unauthorized access to proprietary training data, particularly concerning its earlier R1 model.
Allegations also point to DeepSeek’s controversial use of distillation—a practice where AI models learn by mimicking the outputs of other systems. OpenAI suspects that this technique may have been used to replicate aspects of its proprietary models, raising further concerns over intellectual property theft, according to Associated Press reports.
Further amplifying concerns, DeepSeek’s founder, Liang Wenfeng, met with Chinese President Xi Jinping earlier this year, raising additional worries about the company’s ties to Chinese state policies.
DeepSeek has also faced logistical challenges in its development. The company has been stockpiling Nvidia chips amid concerns that U.S. sanctions could limit its access to advanced AI hardware. Without cutting-edge hardware, DeepSeek may struggle to keep pace with its global competitors.
To continuing strengthening of US export controls on AI chips has also led to stockpiling of Nividia chips by Chinese companies who want to run models like those from DeepSeek.
OpenAI’s Regulatory Strategy and Policy Push
OpenAI’s proposal isn’t solely focused on banning DeepSeek. The company is also lobbying for broader AI regulations aimed at strengthening U.S. technological leadership.
Key recommendations include increasing investment in AI infrastructure, easing restrictive state laws, and implementing tighter export controls to prevent advanced AI technologies from reaching adversarial nations.
Additionally, OpenAI advocates for allowing AI models to train on copyrighted materials under the fair use doctrine. This policy, according to OpenAI, is essential to maintaining the competitiveness of U.S.-developed AI systems.
Without it, OpenAI argues, U.S. companies could fall behind in the global AI race, with a potential loss of $175 billion in investment shifting towards Chinese companies.
