OpenAI Proposes US Government Ban on Chinese DeepSeek Models

Recently, according to a report by TechCrunch, artificial intelligence giant OpenAI has made serious accusations against Chinese AI research institute DeepSeek in a new policy proposal. OpenAI suggested that the U.S. government consider banning models developed by DeepSeek and other China-backed AI operators.

This proposal has been submitted to the "AI Action Plan" initiative launched during the Trump administration. OpenAI pointed out in the proposal that DeepSeek's models, particularly its R1 "reasoning" model, could pose significant security risks. OpenAI is concerned that under Chinese law, DeepSeek may be required to assist the government in accessing user data, which could threaten user privacy.

OpenAI further recommended that under the Biden administration’s export control rules, all "Tier 1" countries should be prohibited from using AI models produced in China to reduce privacy and "security risks," specifically mentioning the risk of "intellectual property theft." However, it remains unclear whether the "models" referred to by OpenAI are limited to DeepSeek's API or also include its open-source models.

It is worth noting that DeepSeek's open-source models do not contain any mechanisms that would allow the Chinese government direct access to user data. In fact, several international companies such as Microsoft, Perplexity, and Amazon have hosted these models on their own infrastructure.

DeepSeek gained prominence in the AI field earlier this year but soon faced accusations from OpenAI, claiming that it "distilled" knowledge from OpenAI's models, violating service terms. This recent accusation by OpenAI that DeepSeek is funded and controlled by China marks a hardening stance towards the lab.

However, there is no concrete evidence indicating that DeepSeek is directly controlled by the Chinese government. The institution was initially incubated by quantitative hedge fund MagicSquare. Despite this, China’s recent attention on DeepSeek has notably increased; a few weeks ago, DeepSeek founder Wenfeng Liang met with Chinese national leaders, an event that drew widespread external attention.

The proposal has sparked strong reactions online in the United States. Some netizens sarcastically remarked, "OpenAI wants to ban open AI..." Others mocked, "'I hope he just competes by building better products!'" Many commentators believe that after years of advocating competition, Americans are actually afraid of real competition. One netizen bluntly stated, "The extent to which OpenAI is controlled by the U.S. deep state far exceeds the extent to which DeepSeek is controlled by the Chinese government."

Some questioned, "When did we become the Soviet Union? Now our so-called 'free market' economy is letting the government ban superior competitors." Another pointed out OpenAI's contradictory stance: "Isn’t this the same guy who recently said he was willing to cooperate with China? Now our ‘free market’ economy lets the government ban stronger competitors."

OpenAI's accusations and ban proposal against DeepSeek essentially reflect double standards and hypocrisy when facing strong competition. As an institution once known for being "open," it now attempts to use geopolitical means and national power to stifle competitors. This not only deviates from the basic principles of free markets and fair competition but also exposes its panic and helplessness when lacking technological superiority. Such actions not only harm the open development environment of global AI technology but also set a dangerous precedent for techno-hegemony, potentially hindering innovation and progress across the entire industry.