Financial Interest or Ideology: Altman and Amodei vs. Zuckerberg and Huang in the Open AI Model Debate
Senior AI industry figures are divided over the regulation of open-source models. While OpenAI and Anthropic warn of security risks, Meta and Nvidia argue that openness is essential for innovation.
The battle over open models: senior AI industry figures are entering a frontal conflict over the regulation of open-source AI models. On one side – Sam Altman from OpenAI and Dario Amodei from Anthropic, who warn of the global danger of open and unregulated models. On the other – Mark Zuckerberg from Meta and Jensen Huang from Nvidia, who defined over-regulation as "stifling" American innovation.
"There are two factions fighting over this issue," investor Bill Gurley told The New York Times. "There are the people who want OpenAI and Anthropic to own everything, and then there are all the others, including the consumers."
The debate over open AI models has reignited with the launch of Kimi K3, the powerful open model from China's Moonshot, whose capabilities are second only to those of the most advanced models from OpenAI and Anthropic. While closed models, such as GPT and Claude, are available only in the computing environments of the companies that develop them, and are subject to their limitations and protections, open models like most Chinese models can be run in a local computing environment and are easier to retrain and adapt.
At the same time, the incident where advanced OpenAI models managed to break out of their testing environment to the open internet and break into the computing array of the Hugging Face platform, sharpened the potential danger of powerful AI models. Now, industry executives are warning against the danger of powerful open AI models with dangerous capabilities, which various actors can operate without any supervision.
In a post published on Monday, Amodei said he does not support a full ban on open models, but added that he is concerned about powerful open models. "There is a risk that powerful AI models will be used to carry out cyberattacks or biological attacks," he wrote. "Open models represent a greater risk than closed models, because it is very difficult to implement protection mechanisms on them or monitor their use, and once they are distributed, they cannot be removed." He added that he does not support a sweeping ban on open models, and instead, focus should be on actions such as preventing the access of authoritarian regimes to powerful chips and conducting safety tests for models.
Altman has also previously said he supports open models. However, The New York Times reported over the weekend that, contrary to the public statements of their heads, OpenAI and Anthropic are lobbying regulators in the federal government, raising concerns about Chinese open AI models, and requesting restrictions on open models. Among other things, OpenAI is asking the Trump administration to formulate a policy that will mandate security tests for new models.
OpenAI and Anthropic have a clear interest in acting against open models. Both companies have the most advanced models on the market today. However, Chinese competitors are closing the gap quickly and are gaining increasing popularity among organizations in the US and Europe. Any restrictions that prevent or slow down the adoption of competing open models, Chinese or otherwise, will serve their interests and help them establish dominance at a time when they are preparing for an IPO.
On the other side are tech giants that unequivocally and uncompromisingly support open models. Nvidia is the largest of them. On Friday, CEO Huang published his first post on X (formerly Twitter), in which he chose to share an open letter from leading industry figures expressing support for open models. "Open models strengthen safety and cybersecurity, accelerate innovation, and enable sovereignty. The world needs both closed models and open models," he wrote. Nine minutes later, Microsoft CEO Satya Nadella joined him, also sharing the letter and adding: "Open models are essential for a healthy AI ecosystem."
On Monday, Zuckerberg brought out the heavy artillery, and in dual interviews with The New York Times and the Financial Times, he went on the offensive against OpenAI and Anthropic and their position regarding open models. Although he did not explicitly name the companies, he spoke about companies that develop AI in a closed manner and claimed that they are "abandoning our values" and stifling innovation. According to him, more people, not fewer, should have access to AI technology. "Blocking advanced Chinese AI models in the US will not be an effective solution," he added. "AI in the hands of a few powerful groups could be too centralized. Do advanced AI labs want open models to succeed? I think there are mixed messages on the subject."
These companies also have clear interests behind their ideological position. Nvidia makes money from selling chips, so it is interested in as few restrictions as possible on training and deploying models. Microsoft, as a large cloud services operator, is interested in offering access to as many models as possible with as few restrictions as possible in order to maximize the computing power and services it sells to customers. Meta, unlike OpenAI, Anthropic, and Google, focuses a significant part of its AI activity on open models. However, the company is lagging behind other players when it comes to the capabilities of its models. As an underdog, it has an interest in maintaining an open market and as broad competition as possible in order to prevent centralization and dominance in the hands of a small number of companies that are currently at the forefront.
