Hinton, Li, Ng defend open AI models amid safety concerns

On August 12, 2026, at the Ai4 conference in Las Vegas, Geoffrey Hinton, Fei-Fei Li and Andrew Ng argued that AI development should stay open, even as they split over how to manage the risks. Open-weight models, which release trained parameters publicly, have become a contentious issue as major labs push for tighter controls.

Hinton, a Nobel Prize winner, distinguished open-source code from open-weight models, warning that open weights make it easy to fine-tune expensive foundation models for harmful purposes such as cyberattacks. Ng, co-founder of Coursera, warned that if a few companies become gatekeepers, innovation could stall, likening the risk to Apple and Google's control over mobile operating systems.

Ng called AI a source of soft power. He cautioned that China's cost-efficient open-weight models could gain adoption in Africa and the developing world, influencing billions of people's views on democracy and human rights. He urged American competitiveness in open-source AI, saying lobbying and fear-mongering in the U.S. are making it harder to compete with Chinese models.

Li, CEO and co-founder of World Labs, rejected a binary choice between open and closed systems. She cited nuclear physics, where papers are open but uranium is regulated, and the Human Genome Project, which created a platform for both profit and scientific progress. "We need some levels of openness," Li said, "but we also will accept closed-source systems."

Hinton acknowledged that open-weight models are permanent: "I think that battle's been lost." He still supports regulation, saying it will help develop AI in a direction that helps people, and added, "You can't leave it to people like Elon Musk and Mark Zuckerberg to decide how AI should be done."

The researchers agreed that nuanced regulation, not blanket openness or closure, is the likely path forward.