The global debate over artificial intelligence has entered a new phase. The question is no longer simply whether governments should regulate AI. It is increasingly about who gets to write the rules that everyone else may eventually follow.
Reports that OpenAI, Anthropic and Google are discussing the creation of a new AI standards body have added a powerful new dimension to that debate. The proposal, associated with Google DeepMind chief Demis Hassabis, would reportedly resemble the US Financial Industry Regulatory Authority, with the major AI companies participating in a system for testing and setting safety standards.
But describing this simply as a plan for a “global AI regulator” would be premature.
The emerging picture is more complicated. The United States is simultaneously promoting international AI standards, resisting some forms of supranational regulation, encouraging American technological leadership and supporting a national framework that could become a model internationally.
That raises a bigger question:
Does Washington want a genuinely global AI regulatory system — or does it want a global AI rulebook built around American technology, standards and institutions?
The New AI Standards Battle Is Already Underway
The reported discussions between OpenAI, Anthropic and Google are significant because these companies sit at the centre of the world’s frontier-AI industry.
According to reporting on the discussions, the proposed body would be self-regulatory rather than a conventional government regulator. It could test advanced systems, establish technical expectations and provide a common framework for evaluating the increasingly powerful models being developed by major laboratories.
That distinction matters.
A government-created global regulator would require agreement among states, legal authority and some form of international enforcement. A company-led standards body would operate very differently.
It could establish technical norms before governments reach political agreement.
And that could give the companies building AI an unusually influential role in determining what “safe AI” actually means.
Why the US Is Suddenly Interested in Global AI Rules
There is clear evidence that parts of the American AI establishment increasingly see international standards as strategically important.
OpenAI wrote in July that American state and federal efforts could lay the groundwork for a US-led global AI framework. The company described the emerging national framework as a potential foundation for international cooperation on frontier AI governance.
That language is important.
It does not describe a neutral global system emerging independently of national competition. It describes American domestic standards becoming a possible foundation for international rules.
The US government’s own AI strategy is similarly explicit about the connection between technological leadership and international standards. The official US AI strategy states that the country that builds the largest AI ecosystem will be positioned to set global standards and capture economic and security benefits.
In other words, Washington increasingly treats AI governance as part of the broader global competition for technological influence.
Trump’s Position Makes the Situation More Complicated
There is, however, an important contradiction inside the American approach.
The Trump administration has repeatedly emphasized innovation and warned against what it describes as excessive regulation. A June 2026 executive order stated that the United States should avoid overly burdensome AI regulation while accelerating innovation and adoption.
At the September 2026 G20 Innovation Ministerial, the United States also promoted the “Carolina Principles,” which favour innovation, research, commercialization and trusted technology adoption. The meeting produced consensus around principles involving AI and standards rather than a binding international regulatory authority.
And on September 15, House Speaker Mike Johnson rejected the idea of a transnational AI regulatory body, arguing that stronger restrictions could undermine the US position relative to China.
So the US position is not simply “America wants a global AI regulator.”
It is closer to a more complex formula:
global standards, American leadership, national regulation where necessary, and resistance to international rules that Washington sees as constraining US competitiveness.
The OpenAI Proposal Reveals the Bigger Question
The proposed industry body could become important precisely because governments are struggling to move at the speed of AI development.
OpenAI has recently changed its position on regulation by openly supporting mandatory national AI safety requirements in the United States. The company has called for capability-based rules, independent assessments, cybersecurity safeguards and incident reporting for advanced systems. It has also argued that US regulation could provide the foundation for international AI safety efforts.
This is a significant shift in the debate.
The industry is no longer speaking only about voluntary corporate ethics.
Some major AI companies are now supporting mandatory standards, external evaluation and stronger government involvement.
Yet the proposed industry body remains industry-led.
That creates an unresolved institutional question:
Who regulates the regulators?
Can AI Companies Be Trusted to Regulate Themselves?
The idea has obvious practical advantages.
Companies such as OpenAI, Anthropic and Google possess technical expertise that many governments do not. They understand model architecture, training systems, security vulnerabilities and emerging capabilities in much greater detail than most policymakers.
A specialist organisation could therefore conduct highly technical evaluations faster than traditional government agencies.
But self-regulation also creates a potential conflict of interest.
The same companies developing frontier systems would participate in defining the standards against which frontier systems are assessed.
That does not prove that a company-led standards system would be ineffective. But it raises legitimate questions about independence, transparency, access to testing data and enforcement.
Those questions become particularly important as AI systems become economically valuable and strategically important.
The World Already Has a Different Model Emerging
The United Nations is moving along a separate path.
In 2026, the UN established an Independent International Scientific Panel on AI with 40 members representing all five UN regions. The panel includes experts from academia, the private sector, civil society and the technical community.
The UN’s Global Dialogue on AI Governance also held its first session in Geneva in July 2026, with every country given a place in the discussion.
But the UN scientific panel is explicitly not a regulator. It cannot impose rules or enforce international standards. Its role is to provide scientific and policy analysis to governments and other stakeholders.
This creates a fascinating contrast.
The UN model emphasizes broad international participation and scientific assessment.
The emerging US technology-industry model emphasizes technical expertise and self-regulation.
The unresolved question is which approach will have greater influence over the rules applied to the most powerful AI systems.
Europe Has Taken a Different Regulatory Route
The European Union has already pursued a more formal regulatory model through its AI legislation.
That makes the transatlantic debate particularly important.
Europe has generally approached AI through legislation, institutional oversight and legally defined obligations, while the US debate has placed greater emphasis on innovation, standards and competitiveness.
The US has also criticized restrictive approaches to technology regulation at international forums. At the September G20 meeting, American officials promoted a relatively light regulatory philosophy and warned against unnecessary restrictions on rapidly evolving technologies.
This means the emerging AI governance system may not become a single global model.
Instead, the world could develop competing regulatory ecosystems.
One could be more rules-based and government-led.
Another could emphasize industry standards and technological innovation.
A third could emerge around China’s own technological and regulatory model.
The China Factor Is Driving Everything
No discussion of US AI regulation can be separated from the American technological competition with China.
US officials have repeatedly presented AI leadership as an economic and national-security priority. The official US AI strategy explicitly links AI dominance with the ability to set global standards.
At the same time, Chinese AI developers are rapidly advancing, narrowing the technological gap with US companies in several areas. Recent reporting has highlighted the growth of Chinese models from firms including DeepSeek, Moonshot AI and Alibaba.
This creates an obvious strategic dilemma.
Washington wants enough regulation to reduce catastrophic and security risks.
But it also wants to ensure that regulation does not slow American companies relative to China.
That explains why proposals for international governance generate such intense debate.
A global regulatory framework could reduce safety risks.
But a poorly designed framework could also restrict the companies the US wants to remain globally competitive.
Is This Really a Global Regulator — Or a Global Standards Machine?
The distinction between regulation and standards is crucial.
Regulation normally involves governments creating legally enforceable requirements.
Standards can be technical specifications, testing protocols, evaluation methodologies or industry practices that governments may later adopt.
A powerful standards organisation can therefore influence regulation without formally becoming a regulator.
This may be what makes the current developments so important.
If OpenAI, Anthropic and Google help create widely accepted safety-testing methods, those methods could eventually influence governments in the United States, Europe, Asia and elsewhere.
The body would not necessarily possess legal authority.
But its standards could acquire enormous practical authority.
That would make the creation of such an organisation a major geopolitical event.
America May Want Global Rules Without a Global Government
This may be the most revealing way to understand the current US approach.
Washington does not appear to be moving toward a single global AI government.
Instead, the American strategy increasingly points toward global coordination built around national leadership and technical standards.
OpenAI’s own policy statements describe US domestic regulation as a potential foundation for a broader international framework.
The US government meanwhile promotes standards, international diplomacy and technological partnerships through its AI strategy.
This approach could allow Washington to influence global AI governance without accepting a supranational regulator capable of overriding US policy.
For the United States, that distinction could be strategically important.
The Biggest Risk Is Regulatory Capture
There is another issue that cannot be ignored.
If the companies building the world’s most powerful AI systems also become the primary architects of the industry’s safety standards, critics could argue that the system risks becoming a form of regulatory capture.
Some technology analysts have already raised concerns that cooperation among dominant AI companies could unintentionally consolidate their position by making compliance standards easier for large firms than smaller competitors.
That is only one interpretation, and it should not be treated as an established fact.
But it highlights a central governance challenge.
AI regulation needs technical expertise.
Yet too much dependence on the companies being regulated can undermine independence.
The solution may require governments, independent auditors, academics, civil society and international institutions to participate alongside industry rather than leaving standard-setting entirely to AI laboratories.
What Would a Genuine Global AI Regulator Need?
A genuine global AI regulatory authority would require much more than an industry standards body.
It would need internationally agreed rules.
It would need mechanisms for independent testing.
It would need transparency requirements.
It would need procedures for handling dangerous incidents.
It would need cross-border enforcement.
And it would need participation from countries outside the US technology ecosystem.
That last point is particularly important.
AI is not only an American technology issue. European governments, Asian economies, African states, Gulf countries and other regions will all face the consequences of increasingly powerful AI systems.
A governance structure designed mainly by a small group of US companies could therefore face questions about legitimacy and representation.
The Real Battle Is Over Who Writes the Future
The debate over AI regulation is increasingly becoming a struggle over institutional power.
Who defines a dangerous AI system?
Who decides when testing is sufficient?
Who verifies corporate safety claims?
Who controls international technical standards?
Who decides what information companies must disclose?
And who has the authority to intervene when an AI system causes serious harm?
These questions are larger than OpenAI, Anthropic or Google.
They concern the future architecture of technological governance.
The United States clearly wants to remain central to that architecture. Its official AI strategy explicitly links global standards to American technological leadership, while OpenAI has described a possible US-led global framework.
But that does not automatically mean Washington wants a traditional global regulatory body.
The current evidence points toward something more nuanced: a US-centred system of standards, national rules, industry cooperation and international diplomacy.
The Future May Be a Global AI Rulebook — Without a Global AI Regulator
The world may ultimately develop international AI standards without creating a single international AI regulator.
That could be a practical compromise.
Governments could retain legal authority, independent bodies could conduct evaluations and companies could contribute technical expertise.
The UN could provide a global forum.
National governments could write enforceable laws.
And international standards organisations could create common technical frameworks.
Such a system would be complicated, but complexity may be unavoidable because AI itself crosses borders.
The technology does not stop at customs checkpoints.
Neither do AI-generated misinformation, cyber risks, automated decision-making systems or the economic disruption caused by increasingly capable models.
The US Is Fighting for Influence Over AI Governance
So, does the United States want a global regulatory body for AI?
The evidence available in September 2026 does not support a simple yes.
There is growing US and industry support for stronger AI safety standards, international coordination and common testing frameworks. OpenAI, Anthropic and Google are discussing a new industry safety body, while OpenAI explicitly describes a possible US-led global AI framework.
At the same time, important US political figures are opposing a transnational regulator, and the Trump administration continues to emphasize rapid innovation and American technological leadership.
The more accurate description is therefore not a straightforward American push for a world AI regulator.
It is a contest over who establishes the standards that could become the foundation of global AI governance.
And that may ultimately be even more important.
Because whoever helps write the technical rules of artificial intelligence may gain influence over how the world’s most powerful technology is developed, tested and deployed for decades to come.
