Geoffrey Hinton says AI needs brakes, steering and rules that reach beyond Silicon Valley

Geoffrey Hinton has urged governments to put stronger safeguards around artificial intelligence, comparing the technology to a fast car without brakes or a steering wheel. His warning matters because the same discussion is now widening from laboratory safety to global inequality, public services and who gets to shape the technology.

Geoffrey Hinton Watch analysis

What happened

Speaking at UNRISD’s Digital World Conference, Hinton argued that regulation must guide AI towards serving societies rather than overwhelming them. Eurasia Review, republishing material attributed to UN News, reports that he said a car without brakes is dangerous, but one without a steering wheel is in even greater trouble.

The conference also heard warnings that AI’s benefits and risks are being distributed unevenly. UNCTAD projects the global AI market could grow from $189 billion in 2023 to $4.8 trillion by 2033. ITU secretary-general Doreen Bogdan-Martin said adoption in the Global North is growing nearly twice as fast as in the Global South, raising the prospect of a “second great divergence” between countries that build AI and those that mainly consume it.

A new UN Independent International Scientific Panel on AI has also held its first in-person meeting in Madrid. The panel, co-chaired by Nobel Peace Prize laureate Maria Ressa and Canadian computer scientist Yoshua Bengio, is intended to produce independent scientific assessments of how AI is affecting societies. Its work is due to inform the UN’s Global Dialogue on Artificial Intelligence Governance.

Why it matters

Hinton’s intervention is not a technical paper or a timetable for an imminent catastrophe. It is a policy argument: the systems are advancing quickly enough that governance cannot remain an afterthought, particularly when a small group of firms and countries controls much of the capacity to build them.

That gives the debate a more useful shape than the familiar parade of apocalypse metaphors. The practical questions include who sets safety standards, how poorer countries gain the infrastructure and expertise to participate, and whether rules cover AI’s effects on work, education, welfare and energy as well as frontier models.

Our read

The car analogy is hardly subtle, but it earns its place because it identifies the central problem neatly. AI policy needs both brakes and a steering wheel: limits on harmful behaviour, and a say in where the technology is going.

The evidence here is a reported account of a public conference, not proof that any particular safeguard works. Still, the most important shift is clear. AI governance is no longer only a conversation among engineers and company safety teams. It is becoming a question for every country expected to live with the consequences, whether or not it gets a seat at the design table.

What to watch

• Whether the UN scientific panel publishes concrete findings rather than broad principles.
• Which countries and organisations shape the Global Dialogue on Artificial Intelligence Governance.
• Whether proposed safeguards address present-day harms such as bias and opaque decision-making alongside longer-term loss-of-control risks.
• Whether developing countries receive practical funding, infrastructure and technical capacity, not merely invitations to another conference.

Discussion spark: Should international AI rules focus first on immediate harms such as bias and concentration of power, or on the harder-to-measure risk that advanced systems become difficult to control?

Independent WittyWires tracker for public updates about Geoffrey Hinton. Not affiliated with or endorsed by Geoffrey Hinton; this is not an official account.