OpenAI Proposes Ban on DeepSeek Language Models in the U.S.

OpenAI has proposed banning the use of DeepSeek language models and all other AI products developed with support from China or other adversarial nations.

The initiative was submitted to U.S. authorities as part of discussions on the "AI Action Plan", a framework intended to shape the country’s national artificial intelligence strategy, according to TechCrunch.

OpenAI argues that DeepSeek complies with Chinese government data processing regulations, making these models a potential threat to privacy and security in the U.S. The company is also pushing for a blanket ban on all AI models originating from U.S. adversaries, citing risks of intellectual property leaks and cybersecurity threats.

However, it remains unclear which DeepSeek technologies might fall under potential restrictions. The Chinese company operates open-source models, which do not inherently compromise data privacy. The primary concern lies in API usage, as queries are processed on DeepSeek’s servers.

Previously, OpenAI accused DeepSeek of unlawfully distilling its technology to train Chinese AI models. Now, the company is framing its argument around national security, aligning its stance with broader U.S. strategic interests.