Demis Hassabis, chief executive of Google DeepMind, has publicly advocated for the creation of a global AI safety organisation led by the United States, according to reports from The Verge and India Today. The proposal represents a notable departure from the tech industry’s historically resistant stance towards regulatory oversight.
Speaking at recent forums, Hassabis suggested that such a body would coordinate international efforts to monitor advanced AI development, establish safety standards, and manage potential risks from increasingly capable systems. The proposal comes as frontier AI labs face mounting pressure from governments worldwide to demonstrate responsible development practices.
The timing is significant. Google DeepMind currently operates at the forefront of AI research, having developed systems including AlphaFold for protein structure prediction and the Gemini family of large language models. Hassabis’s call for oversight from within this position of technical leadership lends considerable weight to the proposal.
The suggestion of US leadership in such a framework reflects geopolitical realities in AI development. American companies—including Google, OpenAI, Anthropic, and Meta—currently dominate frontier model development, whilst the US government has already established preliminary oversight mechanisms through executive orders and the AI Safety Institute.
However, the proposal raises immediate questions about international participation and enforcement mechanisms. China, which has developed its own advanced AI capabilities through companies including Baidu and Alibaba, would likely resist a US-led framework. The European Union has already implemented the AI Act, its own comprehensive regulatory approach that took effect in 2024.
From a business perspective, the implications cut multiple ways. Established AI labs might benefit from regulatory frameworks that impose compliance costs favouring larger organisations with dedicated safety teams. Smaller competitors and startups could face barriers to entry if international standards require extensive testing and documentation before deployment.
Conversely, clear international standards could reduce regulatory fragmentation, allowing companies to develop products for a global market rather than navigating conflicting national requirements. This would particularly benefit cloud providers like Google Cloud, Microsoft Azure, and Amazon Web Services that deploy AI services across jurisdictions.
The proposal also signals a strategic shift in how leading AI companies approach public discourse around safety. Rather than positioning regulation as an external constraint, Hassabis’s comments suggest an attempt to shape the regulatory environment proactively. This follows similar moves by OpenAI CEO Sam Altman, who testified before the US Congress in 2023 calling for AI oversight.
Critics may view such proposals as attempts to establish regulatory capture, where incumbents influence rules to their advantage. Academic researchers and civil society organisations have consistently argued for independent oversight bodies free from industry influence.
The practical challenges remain substantial. Any global watchdog would need to address questions of funding, technical expertise, enforcement authority, and political independence. The International Atomic Energy Agency, often cited as a potential model, took years to establish and operates with significantly more limited scope than what comprehensive AI oversight would require.
Market observers should monitor several developments in coming months: responses from other major AI labs, particularly OpenAI and Anthropic; reactions from Chinese technology companies and government officials; and whether the proposal gains traction in international forums such as the G7 or United Nations.
Hassabis’s intervention marks a watershed moment in AI governance discussions, moving the conversation from whether international oversight is needed to what form it should take. Whether this translates into concrete institutional arrangements will depend on diplomatic negotiations that extend far beyond the technology sector itself.
