LinkedIn cofounder and billionaire venture capitalist Reid Hoffman, Instacart CEO Fidji Simo, Hugging Face cofounder and CEO Clement Delangue, and Arthur Mensch, cofounder and CEO of the prominent Paris-based AI startup Mistral, signed the letter. British entrepreneur and venture capitalist Brent Hoberman, who runs Founders Forum, and Ren Ito, the cofounder and chief operating officer of Japan-based Sakana AI, were also among the 10 initial signatories.

“The choices made in the development and deployment of AI systems impact our societies in profound ways,” the letter states. “These decisions can ensure AI works natively in multiple languages, contributes to the creation of life-saving drugs, and improves educational outcomes. Or they can fall short.” It goes on to say that “to achieve the best out of AI, society must be in charge.”

The letter says that “in practice” the creation of public goods in AI means creating what it terms “high-value data sets” that are publicly accessible in “privacy-preserving and safe ways.” The letter does not specify exactly what kinds of information would be contained in these data sets but it implies that at least some of the data will biological or medical data.

“These resources will enable non-profits and researchers to make breakthroughs in life-threatening diseases such as cancer while also creating opportunities for start-ups to thrive using these shared tools,” the letter states.

The letter also calls for the creation of incentives for “the development of smaller, open AI models that cater to people’s needs and are more environmentally friendly.” It says that open-source AI should also be made more transparent, safe and accessible.

Delangue, whose company Hugging Face acts as a kind of marketplace of AI models, is a vocal advocate for open source AI. Mensch’s company Mistral offers a variety of open-source and commercially-licensable models, as well as proprietary models that can only be accessed through a paid application programming interface.

The letter also calls for further research into AI’s societal impacts. “We need a deeper understanding of how AI is used in critical sectors like education, healthcare, and homecare, fully grasping both its positive and negative impacts on individuals and communities,” it states.

Although the AI Action Summit is a successor to two previous meetings of global leaders on AI governance held at Bletchley Park, England, in November 2023 and Seoul, South Korea, in May 2024, both of those previous summits focused primarily on ways to mitigate AI’s potential existential risk—the idea that a powerful AI system might intentionally or inadvertently kill off all humans—and its potential for other catastrophic harms, such as helping a terrorist group engineer a potent bioweapon or helping hackers to engineer a devastating malware worm.

The French government, however, has been eager to shift the focus of the discussions to a broader set of both risks and benefits from AI. These include the scientific and economic benefits of AI, as well as how AI could have a positive impact on society and culture. It also wants to discuss a broader set of risks, including the dangers of job displacement and algorithmic discrimination, alongside discussions of potentially existential risks.

Correction, Feb. 10: An earlier version of this story erroneously identified Ren Ito as Sakana’s CEO. He is the company’s chief operating officer.