IGF Mauritius

Internet Governance Forum Mauritius

Cybersecurity Data Governance Digital Trade Generative AI Geopolitics of IG

OpenAI’s manifesto: Neo-mercantilism Enters Global AI Governance

It was probably inevitable, given the current world situation. In a filing with the US Office of Science and Technology Policy for the Trump administration’s “AI Action Plan,” OpenAI played the geopolitical card, urging the U.S. government to ban its Chinese competitor. In what is becoming an increasingly tiresome trope, now in its third iteration after Huawei and TikTok, it states: 

DeepSeek could be compelled by the CCP to manipulate its models to cause harm. And because DeepSeek is simultaneously state-subsidized, state-controlled, and freely available, the cost to its users is their privacy and security, as DeepSeek faces requirements under Chinese law to comply with demands for user data and uses it to train more capable systems for the CCP’s use. 

The company calls for banning the use of “PRC-produced” models in all countries considered “Tier 1” under the Biden administration’s export rules. The OpenAI filing tries to cloak its protectionist policies in democratic garb, calling for a fragmentation of the global market along the lines of a Cold War alliance regime:  

A comprehensive export control strategy should do more than restrict the flow of AI technologies to the PRC—it should ensure that America is “winning diffusion”, i.e., that as much of the world as possible is aligned to democratic values and building on democratic infrastructure. 

By “alignment with democratic values,” OpenAI means “using our product.” It also means pushing other nations to “deploy AI in line with democratic principles set out by the US government.” (our emphasis) 

This is not, of course, a liberal-democratic strategy. It is a neo-mercantilist strategy that fuses trade and economic development with an attempt to enhance the political power and security of the U.S. state. Indeed, the statement opens with a quote from a Trump Executive Order saying, “It is the policy of the United States to sustain and enhance America’s global dominance [in AI] ….” OpenAI wants to become a U.S. national champion, promoted and protected by the state.

The Slippery Slope

As we have pointed out repeatedly during the debate over the TikTok ban, these attempts to portray any digital interaction with Chinese companies or applications as a national security threat is a recipe for the politicization, regulation and de-globalization of all digital technology, from the little WiFi router in your home, to laptops, to the thousands of software applications on your phone, to websites, and the digitized automobiles that will eventually make up a big chunk of all world trade. Geo-politicizing digital technology in this way is a slippery slope that will lead to extensive restrictions on everything ICT. Ironically, OpenAI uses the China threat to argue against the 780+ state-level laws and regulations being piled on AI. This welter of local jurisdictional regulation, they argue, will hamper American firms in their “race” with China. But as Microsoft has learned, efforts to turn machine learning applications into a geopolitical race will lead to lots of restrictions and regulations that harm market development. 

To elevate the Chinese threat posed by DeepSeek, OpenAI had to make false statements or obscure important distinctions.  DeepSeek is not state-subsidized, and its use is not state-controlled. OpenAI conflates DeepSeek’s model, which users can download and modify, with its platform for distributing the app, which is based in China.

The tired old argument that China’s National Intelligence Law means that every Chinese company turns over all of its data to the CCP, and that all this data is of such value that it poses a national security threat to the U.S. overlooks the simple fact that no major data breach or compromise of US national security attributed to China has ever relied on Chinese-owned apps or equipment.  

In an annoying rhetorical turn typical of the tech industry, the OpenAI statement wants to have it both ways: preaching freedom and innovation while making a case for market restriction and protectionism. It calls for  

freedom of intelligence, by which we mean the freedom to access and benefit from AGI, protected from both autocratic powers that would take people’s freedoms away, and layers of laws and bureaucracy that would prevent our realizing them.  (emphasis in original)

Yet the company argues that we should have additional layers of laws and bureaucracy, run by a domestic (increasingly autocratic) power, that prevents people from accessing and benefiting from foreign AI applications. The hypocrisy gets even deeper the further you go into their statement. Open AI says “the best future is one in which we move forward with democratic AI.” OpenAI defines “democratic AI” as: 

  • A free market promoting free and fair competition that drives innovation. 
  • Freedom for developers and users to work with and direct our tools as they see fit, in exchange for following clear, common-sense technical standards that help keep AI safe for everyone, and being held accountable when they don’t. 
  • Preventing government use of AI tools to amass power and control their citizens, or to threaten or coerce other states. 

So to achieve democratic AI we must  

  • Restrict the free market and reduce competition if it comes from China 
  • Prevent Americans from exporting goods and services to Chinese developers and consumers 
  • Prevent developers and users from importing Chinese companies’ tools 
  • Ask the U.S. government to control its citizens’ access to applications  

One commenter on the TechCrunch article about the AI filing skewered Open AI’s hypocrisy:  

Effectively, Sam [Altman] is arguing for the US to create its own Great Firewall to block US consumers from accessing open-source code from abroad and running it on their own computers, which is both completely novel and completely nonsensical, even more nonsensical than the attempts at encryption export-bans in the 1990s. I also find it particularly ironic that he argues distilling OpenAI models “violates their IP” when OpenAI has been arguing this whole time that training in violation of IP was “unavoidable” for developing foundation models. 

The post OpenAI’s manifesto: Neo-mercantilism Enters Global AI Governance appeared first on Internet Governance Project.