---
title: "DeepMind's Hassabis calls for a US-led global watchdog for advanced AI"
description: "Demis Hassabis, chief executive of Google DeepMind, has proposed a US-led global body to vet the world's most powerful artificial intelligence models before release, arguing that the pace of progress now demands a systematic approach rather than ad hoc rules."
category: "Technology"
category_url: https://newsparlor.com/category/technology
author: "Chloe Bennett"
published: 2026-07-14T13:34:00.000Z
updated: 2026-07-14T13:34:00.000Z
canonical: https://newsparlor.com/article/hassabis-calls-for-us-led-global-ai-watchdog
tags: ["artificial-intelligence", "ai-regulation", "demis-hassabis", "google-deepmind", "technology-policy"]
---
# DeepMind's Hassabis calls for a US-led global watchdog for advanced AI

Demis Hassabis, chief executive of Google DeepMind, has proposed a US-led global body to vet the world's most powerful artificial intelligence models before release, arguing that the pace of progress now demands a systematic approach rather than ad hoc rules.

The head of one of the world's leading artificial intelligence labs has called for the creation of a global watchdog, led by the United States, to screen the most capable AI systems before they reach the public.

Demis Hassabis, the chief executive of Google DeepMind, set out the proposal in [a personal manifesto published on July 14](https://www.axios.com/2026/07/14/demis-hassabis-ai-regulation-google-deepmind), titled "A Framework for Frontier AI and the Dawning of a New Age." He argued that the time had come for a more "systematic" approach to oversight, funded by industry, staffed by technical experts and answerable to the US government.

## How it would work

Hassabis proposed a body modeled on FINRA, the industry-funded organization that polices Wall Street under government supervision. Under the plan, developers of the most advanced "frontier" models would share them with the watchdog up to 30 days before release for safety testing, probing for dangerous capabilities in areas such as cyberattacks, the design of biological weapons and "deception," meaning a model's ability to conceal its reasoning or manipulate users.

The organization would be run by a majority-independent board that included leading computer scientists alongside representatives of industry, government and the open-source community. Crucially, Hassabis said the rules should apply to all frontier-class models, "no matter their country of origin or whether they are open or closed," with the thresholds updated as the technology advances.

He framed the idea as a response to recent turbulence, describing the US administration's improvised crackdown on a rival lab's models last month as "a bit of a wake-up call" that showed Washington needed something sturdier than one-off directives. Hassabis, who shared a Nobel Prize for his work on protein structure prediction, has previously warned that the industry has only a "precious window" to make advanced systems safe.

## A contested debate

The proposal lands in the middle of an unsettled global argument over how to govern AI. Hassabis and Anthropic's chief executive, Dario Amodei, [called for a US-led coalition on frontier AI standards at a G7 meeting in June](https://www.cnbc.com/2026/06/17/anthropic-amodei-google-hassabis-us-ai-coalition-g7.html), reflecting a view among some developers that a single, credible standard is needed quickly.

Others are wary. Critics of industry-designed oversight argue that heavy pre-release testing requirements could entrench the largest companies and disadvantage smaller and open-source developers that lack the resources to comply. Sceptics also question how a US-anchored body could bind labs operating in other jurisdictions, particularly China.

The European Union, meanwhile, has pursued a different path, with its own AI Act moving toward implementation and officials stressing that legitimacy requires broad, multilateral participation rather than leadership by any one country. Whether governments embrace Hassabis's model, adopt a more international framework, or continue with a patchwork of national rules remains an open question, and his manifesto is a bid to shape that choice rather than a settled outcome.