APREX · EST. 2026
Demis Hassabis has long been one of the most influential voices in the debate over safe AI development. Now he is going a step further: in a recent post, the head of Google DeepMind has come out in favour of establishing an international oversight body for artificial intelligence — one with genuine authority to slow development if frontier models prove too dangerous.
The US should take the lead
According to The Verge, Hassabis argues that the US is the natural anchor for such an initiative. His reasoning rests on the country's dominant position in both technology development and the global economy, which he believes makes it the right actor to set global standards.
Hassabis is not the first to advance the idea of international AI regulation, but it is rare for a sitting CEO of one of the world's most powerful AI companies to explicitly call for a supranational oversight body with the power to apply the brakes.
A sitting AI chief actively calling for a global watchdog with stop-gap authority — that is not an everyday occurrence in this industry.
A fragmented regulatory landscape
Hassabis's intervention lands in an increasingly complex international picture. Several parallel processes are already under way:
The Council of Europe opened its framework convention on artificial intelligence for signature in September 2024 — the first legally binding international treaty on AI. More than 50 countries, including EU member states, have signed on. The convention requires AI systems to respect human rights, democracy, and the rule of law throughout their entire lifecycle, and excludes national security and defence from its scope.
The United Nations, for its part, has recommended an "AI Standards Exchange" in which representatives from standards bodies, technology companies, and civil society would coordinate definitions and requirements across borders. The UN's High-Level Advisory Body published its report "Governing AI for Humanity" in September 2024, concluding that the need for global AI governance is "irrefutable."
The OECD's AI Principles, originally adopted in 2019 and revised in 2024, are regarded as the first intergovernmental standard in the field. These principles have shaped national AI policy around the world and provided the foundation for, among other things, the EU's AI Act.
Who actually gets to decide?
The challenge with Hassabis's proposal is the same one that confronts most international regulatory initiatives: who grants the body legitimacy, what kind of mandate should it hold, and how do you prevent it from becoming an instrument of geopolitical power rather than genuine technical oversight?
The existing structure is fragmented. The Council of Europe and the EU operate with legally binding frameworks. The UN relies on voluntary coordination. The OECD principles are influential but not legally binding. A new US-led body would have to find its place within this landscape — and persuade countries such as China, which is not party to any of the Western-dominated initiatives, to participate.
Genuine concern or lobbying?
It is worth noting that Hassabis is advancing this position as the leader of one of the most resource-rich AI laboratories in the world. The call for regulation can be read as a sincere concern for safety — but also as a strategic move by an established company that already has the resources to meet stringent requirements, unlike smaller players.
The source of the statement is a blog post by Hassabis himself, as reported by The Verge. It is not a scientific analysis or a political agreement — and should be read accordingly.
Nevertheless: the fact that one of the industry's most respected figures is now publicly calling for an international brake mechanism adds weight to the regulatory debate at a moment when pressure for concrete global solutions is greater than ever.