Three leading AI researchers—Geoffrey Hinton, Fei-Fei Li and Andrew Ng—argued at the Ai4 conference that AI should remain broadly accessible despite mounting safety concerns around open-weight models. They warned that allowing a small number of major technology companies to control advanced AI could limit innovation, shape industry rules in their favour and restrict public access to an important technology.

Ng advocated competition among multiple model providers and greater openness to avoid gatekeepers, while Hinton distinguished open-source code from releasing trained model weights. He said open weights can make it cheaper to adapt powerful models for harmful uses such as cyber-attacks, but acknowledged that their widespread availability is now irreversible; he also stressed that AI’s potential benefits for productivity, education and healthcare should be weighed alongside its risks.