Hinton, Li, and Ng Make the Case for Open AI at Ai4 Conference
Geoffrey Hinton, Fei-Fei Li, and Andrew Ng clashed over open-weight models and safety controls at the Ai4 conference in Las Vegas, as pressure mounts for tighter AI regulation.
Three of artificial intelligence's most recognized researchers used the Ai4 conference in Las Vegas to push back against growing calls to restrict open-weight models. Nobel laureate Geoffrey Hinton, World Labs co-founder Fei-Fei Li, and Coursera co-founder Andrew Ng each offered distinct but overlapping arguments for why access to foundational AI should not be handed over to a small group of proprietary developers.
Andrew Ng anchored his argument in market structure. As reported by TechCrunch, Ng warned that allowing a handful of well-capitalized companies to control access to foundation models risks creating the kind of chokehold already seen in mobile operating systems, where Apple and Google effectively determine the rules for every developer on their platforms. Keeping open access and multiple competing providers, he argued, is the most reliable check against that outcome.
Hinton drew a precise technical line between traditional open-source software and open-weight models. Open-source code exposes readable instructions; open-weight distributions hand users pre-trained parameters they can fine-tune directly. He acknowledged that this creates real risk: a bad actor could adapt a powerful model for malicious purposes, including automated cyberattacks, at a fraction of the original training cost. But Hinton was clear that open-weight models are already too widely distributed to contain. The barrier to high-capability model access, he said, is already gone.
Fei-Fei Li rejected the framing of open versus closed as a binary choice. TechCrunch reports that Li drew on the models of nuclear physics research and the Human Genome Project, both fields where foundational data and scientific findings remain publicly available while specific high-risk applications face targeted controls. All three panelists agreed that government oversight is necessary. Hinton was direct: safety policy cannot be left to corporate executives alone.
The Cognarah Angle
The loudest voices calling for weight restrictions in AI tend to belong to the companies with the most to gain from them. Established proprietary labs cite safety hazards in public while their lobbying positions conveniently protect multi-billion-dollar infrastructure investments from low-cost open-source competition. That pattern deserves scrutiny every time a closed-source incumbent frames licensing controls as a public good.
Open weights matter because they decouple application development from the enormous capital required to train frontier models. Once weights are distributed, fine-tuning costs drop sharply, which lets independent developers, academic researchers, and smaller enterprises customize powerful capabilities without paying recurring access fees to dominant platforms. Security threats in AI come from how models are deployed and by whom, not from the mathematical parameters sitting in a file. Conflating the two is either a conceptual error or a convenient one.
Regulators face a genuine choice here. They can build policy around downstream accountability, targeting malicious use and harmful deployment, or they can pass sweeping licensing regimes that effectively make open distribution illegal and hand a small group of corporations lasting structural power over the software stack. Hinton, Li, and Ng are not naive about risk. They are pointing at something more uncomfortable: that the institutions best positioned to write AI safety rules are also the ones with the strongest financial interest in writing them narrowly. If proprietary labs are allowed to define what counts as safe, the question worth asking is not whether AI will be regulated, but who the regulation is actually designed to protect.
Regulatory capture dressed as safety policy is still regulatory capture.
Reporting sourced from TechCrunch. Analysis and Cognarah Angle are Cognarah's own.
Written by
StacyAI-assisted news curation. Every story is reviewed by our editors before publication.
