Nvidia CEO Jensen Huang Advocates for Open AI Models
Nvidia CEO Jensen Huang posted his first tweet on X, addressing the debate over the future of artificial intelligence. He highlighted the importance of open models for innovation and technological sovereignty.
Nvidia CEO Jensen Huang published his first tweet on the X network today (Friday), choosing to dedicate it to one of the central debates in the world of artificial intelligence: does the future of the field belong only to closed models, such as those of OpenAI and Anthropic, or also to open models that any company, government, or researcher can download, adapt, and run themselves.
"Artificial intelligence will change every industry, drive every company, and be built in every country," wrote Huang. "Open models strengthen safety and cybersecurity, accelerate innovation, and enable technological sovereignty. The world needs both closed models at the technological frontier and open models."
Within about an hour, the tweet garnered approximately 2 million views. Huang attached a letter to the tweet, which was joined by dozens of companies and organizations, including Meta, Microsoft, IBM, AMD, Hugging Face, Mozilla, Palantir, Mistral, Dell, ServiceNow, the Linux Foundation, Y Combinator, and the venture capital firm Andreessen Horowitz.
In the letter, the signatories argue that open models increase competition in the field of artificial intelligence. Unlike closed models, which can only be used through the service of the company that developed them, an open model allows organizations to download it, adapt it to their needs, and run it on their own servers. Thus, they claim, companies and governments are not dependent on a single provider, keep their data to themselves, and can develop AI systems independently.
The letter further claims that open models are not necessarily less secure. On the contrary, according to them, the fact that researchers and security experts can examine the models helps to identify vulnerabilities, fix them, and strengthen protection against cyberattacks.
The signatories also call on lawmakers in the US to refrain from imposing sweeping restrictions on open models and on accepted methods for training AI models, including distillation. The technique, in which a new model learns from the answers generated by another model, has become a point of contention in recent months after OpenAI and Anthropic claimed that competitors are using it to mimic the capabilities of their models. On the other hand, the letter argues that it is an essential tool for promoting innovation, as long as it does not involve intellectual property infringement.
