OpenAI Urges US Ban on DeepSeek AI Models

OpenAI has recommended that the US government prohibit AI models from the Chinese lab DeepSeek, citing the project’s state sponsorship and control.

The document was submitted as part of the Trump administration’s “AI Action Plan.” It claims that DeepSeek’s products, including the controversial R1 neural network, are unsafe and comply with Chinese government demands for user data.

A ban on models from China in all countries “Tier One” would prevent privacy and security threats, including the “risk of intellectual property theft,” the American startup asserts.

It remains unclear whether the call applies to open DeepSeek models that do not have mechanisms for collecting and transmitting user data to the Chinese government. Companies like Microsoft, Perplexity, and Amazon host them on their infrastructure.

Previously, OpenAI suspected DeepSeek of information theft. Nonetheless, the company’s head, Sam Altman, described DeepSeek-R1 as “impressive,” considering its capabilities for its price.

The startup also urged the US government to develop an intellectual property strategy that “preserves the ability of American models to train on copyrighted materials.”

“America has so many AI startups, investments, and scientific breakthroughs largely because the fair use doctrine fosters the development of artificial intelligence,” writes OpenAI.

Back in August 2024, the firm announced a partnership with Condé Nast to display content from major publications.