Introduction
In bold and unprecedented remarks reflecting growing concern among tech leaders and senior experts regarding the future trajectory of intelligent and autonomous machines, Demis Hassabis, co-founder and CEO of leading AI research firm Google DeepMind, has issued a direct call for the United States to take the lead in establishing a new, independent global regulatory body. According to a comprehensive report published on Monday, Hassabis demands that this proposed agency be equipped with broad powers to inspect and test the world’s most advanced AI models, as well as the legal and executive authority to immediately halt their development and hold developers accountable if they are classified as an existential or severe threat to societal and human security. This concrete proposal represents a major evolution and radical shift in Hassabis’s stance, moving clearly from general theoretical warnings about autonomous technology to an urgent demand for a specific, rigorous institutional and legal framework to govern the industry. These strong statements come at a sensitive time as intense international debate escalates over how to govern and regulate frontier AI systems, with varying visions and stances among major powers and tech institutions worldwide.
- Building regulatory momentum starting from the G7
- An increasingly crowded and overlapping regulatory governance landscape
- Challenges of aligning global visions and interests
- FAQs
Building regulatory momentum starting from the G7
Demis Hassabis worked diligently and persistently throughout 2026 to lay the political foundations for this important regulatory step. In an influential and forward-looking academic speech delivered at Stanford University last May, Hassabis described rapid and unprecedented developments in artificial intelligence as a “transition and evolution for the human species as a whole,” warning firmly that this critical historical phase leaves humanity with a “frighteningly narrow margin for error” over the coming decade. He drew on these warnings to argue urgently for increased international coordination and close cooperation in AI governance, to be implemented within five to ten years at most to avoid the severe consequences of an uncontrolled tech arms race.
During discussions at the G7 summit held in Évian-les-Bains, France, last month, Hassabis collaborated closely with Anthropic CEO Dario Amodei, jointly advocating for a US-led strategic alliance to establish binding global AI rules. This proposal was presented and discussed in a high-level closed-door meeting personally attended by US President Donald Trump. Following this pivotal meeting, G7 leaders formally pledged to enhance joint cooperation in addressing advanced AI risks, tasking senior financial officials, regulators, and cybersecurity experts with immediately assessing the potential impacts of these advanced models on financial market stability and international labor markets.
An increasingly crowded and overlapping regulatory governance landscape
Hassabis’s bold proposal enters a global regulatory environment that is increasingly crowded with initiatives, laws, and sometimes conflicting proposals. In June, the Trump administration issued an executive order titled “Promoting Advanced AI Innovation and Security,” directing all federal agencies to develop strict benchmark testing to evaluate the true capabilities of advanced AI models, while setting technical thresholds for “covered frontier models.” Notably, this US executive order relies primarily on voluntary cooperation between developers and the federal government rather than mandatory licensing systems, which critics argue is insufficient for deterring non-compliant companies.
Internationally, the United Nations launched the “Independent International Scientific Panel on Artificial Intelligence” in February 2026. Comprising 40 experts worldwide, the panel is tasked with producing evidence-based reports on the risks and opportunities of AI development. The panel presented its first comprehensive annual report during the Global AI Governance Dialogue in Geneva earlier this month, adding another layer to international efforts.
Challenges of aligning global visions and interests
Despite the significance of the proposal, an open question remains as to whether Hassabis’s unified global regulatory body will complement existing frameworks or spark a turf war. More complex is how this new independent body can diplomatically reconcile US leadership and hegemony with the multilateral and collaborative approaches favored by European and developing nations, ensuring fair governance that protects humanity without stifling innovation.
FAQs
Question: What is the main proposal put forward by the Google DeepMind CEO?
Answer: Demis Hassabis proposed establishing a US-led global regulatory body with powers to inspect advanced AI models and suspend their development if they pose severe risks to society.
Question: How does Hassabis view the impact of AI on humanity’s future?
Answer: He views it as a fundamental transition for the human species, warning that the margin for error over the next decade is extremely narrow and requires strict international coordination.
Question: What regulatory actions has the Trump administration taken?
Answer: It issued an executive order directing the development of evaluation tools for advanced models, relying on voluntary cooperation between developers and the government rather than mandatory licensing.
Question: Are there parallel international efforts for deep tech governance?
Answer: Yes, the United Nations launched an independent scientific panel to assess risks and opportunities, alongside ongoing discussions among G7 leaders.
