The U.S. government's sudden decision to block Anthropic from exporting its newest AI models has triggered a geopolitical showdown over who controls access to artificial intelligence technology. Last week, the Commerce Department ordered Anthropic to halt access to its Mythos 5 and Fable 5 models for foreign users, citing national security concerns tied to potential safety guardrail bypasses. This move has exposed a harsh reality for companies and governments worldwide: access to cutting-edge AI can be revoked overnight without detailed explanation, and two competing governance philosophies are now battling for global dominance.

Why Are Nations Suddenly Worried About AI Sovereignty?

The Anthropic ban sparked intense debate at the G7 summit in France, where world leaders confronted a troubling question: what happens when one country controls the technology that underpins critical infrastructure? French President Emmanuel Macron warned that if the U.S. can "turn off the switch from one day to the next," it would harm not only European customers' economic interests but also damage the AI companies themselves. Indian Prime Minister Narendra Modi expressed similar concerns, emphasizing that democratic nations must have unfettered access to top AI models to protect critical infrastructure.

Emmanuel Macron

This recognition reflects a broader shift in how governments view AI dependency. When a country's financial systems, government services, or critical infrastructure depend on foreign AI technology that could be cut off at any moment, its decision-making autonomy becomes severely constrained. For nations lacking the resources to develop cutting-edge AI independently, the choice is particularly difficult: accept dependency on U.S. technology with its associated risks, seek technically inferior alternatives, or invest heavily in domestic capabilities that may take years to mature.

What Are the Two Competing AI Governance Models?

Two distinctly different approaches to AI governance are taking shape globally. The U.S. model emphasizes subscription services and export controls, prioritizing technology sharing among "trusted partners" while excluding China. This approach gives the U.S. government ultimate control over technology access but raises concerns among allies about technological dependency. By contrast, China's model emphasizes openness and inclusivity, offering free or low-cost open-source models that anyone can download and use. Chinese open-weight models like DeepSeek and Qwen are available to anyone with an internet connection, appealing particularly to Global South countries that cannot afford enterprise AI subscriptions.

At the G7 summit, Anthropic CEO Dario Amodei and Google DeepMind CEO Demis Hassabis called for a U.S.-led AI coalition to establish international rules and standards. Amodei proposed that areas of international cooperation should include structured access to frontier AI models, trade in chips and critical components that excludes China, and collaboration to address AI risks in cyber operations, bioterrorism, and intelligence. Meanwhile, Chinese Foreign Minister Wang Yi announced that Beijing is "accelerating the establishment of a global AI cooperation organization" and invited all countries to join, criticizing what he called "closed, exclusive and monopolistic approaches to tech development".

Amodei

How Are Organizations Preparing for Fragmented AI Governance?

As this bifurcation unfolds, organizations and governments are reassessing their AI strategies. The divergence extends beyond access mechanisms to governance philosophy itself. The U.S. and its allies emphasize safety standards, export controls, and value-based partnerships, while China emphasizes multilateralism, technology accessibility, and opposition to what it calls "technological hegemony." Both models are competing for international support, particularly from developing nations that will play an increasingly important role in shaping global AI governance.

To navigate this fragmented landscape, organizations are turning to structured governance frameworks. Two leading approaches have emerged: ISO 42001, the first international standard focused on AI management systems, and the NIST AI Risk Management Framework (NIST AI RMF), a voluntary framework developed by the National Institute of Standards and Technology. These frameworks help organizations manage AI-related risks and ensure responsible deployment, though they differ significantly in scope and implementation.

• ISO 42001 Approach: A certifiable management system standard designed for organizations seeking formal AI governance with defined policies, compliance frameworks, and ongoing risk assessments. It aligns well with industries needing structured governance, such as finance, healthcare, and government-regulated sectors, and integrates with existing ISO standards like ISO 27001 and ISO 9001.
• NIST AI RMF Approach: A flexible, voluntary, and risk-driven framework ideal for organizations prioritizing adaptability in AI risk assessment without needing mandatory certification. It is particularly beneficial for businesses, research institutions, and government agencies that want to manage AI risks while maintaining innovation flexibility.
• Integrated Approach: Organizations can combine both frameworks to create a comprehensive AI management system, using ISO 42001 to establish formal AI policies and NIST AI RMF to conduct ongoing risk assessments and identify emerging AI-related threats.

According to governance experts, organizations looking for a comprehensive AI strategy can integrate both frameworks, leveraging the risk assessment flexibility of NIST AI RMF alongside the structured governance requirements of ISO 42001. This combined approach helps organizations enhance AI trustworthiness by developing formal AI policies and risk-based assessments, aligning AI risk identification with compliance requirements, using risk metrics to strengthen compliance processes, and integrating risk mitigation strategies from both frameworks.

The fragmentation of AI governance creates real challenges for countries and companies that want to engage with both systems. Maintaining access to both U.S. and Chinese AI ecosystems may require navigating contradictory requirements and accepting that some capabilities will remain siloed. This bifurcation could slow innovation, increase costs, and complicate international collaboration on shared challenges like AI safety and ethics.

Meanwhile, industry leaders are pushing for clearer international standards. Aidan Gomez, co-founder of Cohere, noted that "the recent restriction on access to Anthropic's models confirms what we at Cohere have known all along: that companies and democratic nations remaining dependent on a small handful of big tech companies is dangerous to resilience." This perspective has resonated among G7 members and is driving discussions about building alternative AI infrastructure.

The stakes are high. For the first time, AI governance is not merely a technical or ethical concern but a geopolitical battleground where access, control, and sovereignty intersect. Organizations and governments must now choose not just which AI tools to use, but which governance ecosystem to align with, knowing that this choice will shape their technological future for years to come.