Altman proposes US-led international forum for AI safety standards

OpenAI CEO Sam Altman has called for the creation of a US-led international forum to establish global safety standards for AI, arguing that no single country or company should dominate the governance of increasingly capable AI systems.

Writing in an opinion article published in the Financial Times, Altman proposed an international body bringing together governments, independent technical experts, and other stakeholders to develop accepted AI safety standards, provide impartial assessments of AI capabilities and risks, and make advanced AI technologies available to countries and organisations that participate in and comply with agreed rules.

According to Altman, such a forum could also serve as a governance mechanism for frontier AI developers, helping to reduce commercial pressures that may encourage companies to prioritise rapid deployment over safety. He argued that international cooperation has previously enabled countries to manage other strategically important technologies despite geopolitical competition.

To illustrate his proposal, Altman pointed to existing international governance mechanisms such as the International Atomic Energy Agency (IAEA), which oversees the peaceful use of nuclear technology, as well as global aviation safety frameworks and international financial standards. In his view, these models demonstrate that countries can establish common rules for technologies with significant cross-border implications while maintaining national interests.

Altman also argued that the benefits of AI should be shared more broadly, writing that ‘everyone on Earth should benefit from this technology and determine for themselves how best to use it.’ His proposal follows discussions at the recent Group of Seven (G7) summit in France, where executives from OpenAI, Anthropic, and Google DeepMind met with political leaders to discuss international approaches to governing advanced AI models.

A key challenge for any international oversight mechanism, however, remains enforcement. Unlike nuclear facilities or aircraft, frontier AI models are developed within highly secured data centres, making independent verification considerably more difficult. The limited visibility into model training, testing, and deployment has led many experts to question how compliance with international AI standards could be monitored in practice.

Altman’s proposal is not the first call for stronger international oversight of advanced AI. OpenAI and Anthropic have previously supported the idea of international governance mechanisms for frontier AI systems. Anthropic CEO Dario Amodei has argued for a more prescriptive regulatory approach, drawing comparisons with the US Federal Aviation Administration and advocating stronger regulatory oversight for highly capable AI models.

The proposal also comes as governments continue to expand their involvement in AI governance. Alongside national regulatory initiatives, international discussions have accelerated through forums such as the G7, the Global Partnership on AI (GPAI), and the UN.

Earlier this week, the UN’s Independent International Scientific Panel on Artificial Intelligence published its first preliminary assessment of AI opportunities, risks, and governance challenges ahead of the inaugural Global Dialogue on AI Governance in Geneva, reflecting growing international efforts to establish evidence-based approaches to AI governance.

Whether Altman’s proposal develops into a formal international initiative will ultimately depend on governments rather than AI companies. Commenting on broader discussions around AI governance, analysts at the Brookings Institution argued that cooperation between governments and leading AI developers could help establish common standards, but stressed that any future international framework would need effective implementation and enforcement mechanisms rather than relying solely on voluntary commitments.

As governments, international organisations, and AI developers continue debating how to govern increasingly capable AI systems, Altman’s proposal adds to a growing conversation about whether existing institutions are sufficient or whether new international mechanisms will be needed to manage the opportunities and risks associated with frontier AI.

Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!