OpenAI Labels DeepSeek as ‘State-Controlled’ and Urges Ban on PRC-Produced AI Models
In a recent policy proposal, OpenAI has raised significant concerns about the Chinese AI lab DeepSeek, labeling it as “state-subsidized” and “state-controlled.” The proposal urges the U.S. government to consider a potential ban on models produced by DeepSeek and similar operations backed by the People’s Republic of China (PRC).
OpenAI’s Concerns Regarding DeepSeek
The proposal, submitted as part of the Trump administration’s “AI Action Plan” initiative, highlights several critical issues regarding DeepSeek’s AI models, particularly its R1 “reasoning” model. OpenAI claims that these models are insecure due to compliance requirements under Chinese law, which could compel DeepSeek to share user data with the government.
Potential Risks of PRC-Supported Models
OpenAI emphasizes that banning the use of “PRC-produced” models in countries classified as “Tier 1” under the Biden administration’s export regulations could mitigate several risks:
- Privacy Risks: The potential for unauthorized access to user data.
- Security Risks: Vulnerabilities that may expose sensitive information.
- Intellectual Property Theft: The danger of losing proprietary technology and innovations.
Understanding DeepSeek’s Model Offerings
It remains unclear whether OpenAI’s references to “models” pertain to DeepSeek’s API, its open models, or both. Notably, DeepSeek’s open models are hosted by major companies like Microsoft, Perplexity, and Amazon, and are designed without mechanisms that would enable the Chinese government to extract user data.
The Escalation of Tensions
OpenAI has previously accused DeepSeek of violating its terms of service by “distilling” knowledge from its models. However, the latest allegations that DeepSeek operates under direct PRC control signify a notable escalation in OpenAI’s stance against the Chinese AI lab.
DeepSeek’s Connection to the Chinese Government
While a direct link between DeepSeek and the Chinese government is not clearly established, the PRC has shown an increasing interest in the lab over the past few months. Noteworthy was a meeting between DeepSeek’s founder, Liang Wenfeng, and Chinese leader Xi Jinping just weeks ago, which further fuels speculation about the lab’s connections.
For more information on OpenAI’s initiatives and AI policy discussions, visit OpenAI’s official policies or explore related topics on Brookings Institution’s AI research.
