Why Access to Frontier AI Is Becoming a Business Risk

By Rohanshi Vaid

Frontier AI is becoming an important part of national strategy. This is changing the risks faced by businesses that operate across borders.

On 12 June 2026, the US Commerce Department ordered one of the world’s leading AI developers to block foreign nationals from accessing its two most capable models. The company could not reliably distinguish users by nationality. As a result, it switched off both models for all customers worldwide.

The episode showed how quickly government decisions can affect access to advanced AI. Until recently, most business discussions about AI focused on capability. Companies wanted to know what the technology could do and how they could use it.

That question still matters. But businesses must now consider another: who can use the most powerful AI systems, in which countries, and under what conditions?

Access is becoming one of the main government affairs issues surrounding frontier AI.

From capability to government control

Governments increasingly view advanced AI models as strategically important technologies. They are beginning to treat them in much the same way as advanced semiconductors and other technologies with both civilian and security uses.

Frontier models are also developing faster than many governments expected. Their capabilities can support cyber defence, but they can also make sophisticated cyberattacks easier to carry out.

In June 2026, the cybersecurity agencies of the Five Eyes countries issued a joint statement warning that frontier AI could reshape cyber threats in “months, not years.” They called on boards and senior executives to treat cyber resilience as a core business risk.

The message was clear. Governments expect business leaders, not only technical teams, to take responsibility for the risks linked to advanced AI.

This view is now being reflected in export controls, licensing requirements, and restrictions on cross-border access. The June order was especially significant because it applied export control powers, traditionally focused on hardware, directly to AI models.

Why access matters for business

For multinational companies, access is an operational issue.

A model used by teams in one country may not be available in another. Even where it remains available, the provider may offer it on different terms or impose additional conditions.

Access to frontier AI is starting to divide along geopolitical lines. The current US framework places countries into different tiers. Close allies face fewer restrictions, while users in many other countries may require licences.

At the G7 summit, the United States also raised the possibility of giving “trusted partners” greater access to frontier models. This would make access a benefit that governments could grant, limit, or withdraw.

European policymakers responded by renewing calls for “sovereign AI.” Their concern is that Europe should not build important parts of its economy around technology that another government could restrict without warning.

Businesses have also shown greater interest in open-source models and providers outside the United States. Some providers are now promoting their services on the basis that one foreign government cannot order them to switch off access.

This could lead to a divided AI market. A multinational company may use the same business process in several countries but have different tools available in each one.

A fragmented regulatory landscape

Access controls are developing alongside an already fragmented set of AI rules.

Approaches differ widely across the Asia-Pacific region. Some countries have introduced standalone AI laws with binding obligations. Vietnam’s AI Law, which took effect in 2026, is one of the first examples in the region.

Other governments continue to rely on voluntary guidance and industry standards. Several countries have adopted broad legal principles but left regulators to develop the detailed requirements.

The financial sector shows how these rules may develop in practice. Regulators are turning broad AI principles into specific expectations. These include board accountability, records of AI use, human oversight of important decisions, independent model testing, and responsibility for third-party AI.

The Monetary Authority of Singapore, for example, has set out detailed expectations for how financial institutions should manage AI risks.

A company operating in ten markets may therefore need to follow ten different sets of rules. These may overlap, but they are unlikely to be identical.

What this means for global businesses

Three issues deserve attention.

First, companies must treat access as a business planning issue. They can no longer assume that a model will remain available in every market. Model access should form part of procurement, scenario planning, and business continuity discussions. Relying heavily on one provider, or on the rules of one country, creates a clear business risk.

Second, third-party AI risk is now closely linked to government action. A regulator may order a provider to suspend access to a model with little warning. A service agreement may offer some protection, but it cannot prevent the disruption itself.

Third, boards must consider AI and cyber risk together. Regulators and security agencies are increasingly speaking directly to company leaders who control budgets, governance, and risk appetite. Responsibility cannot sit only with technology teams.

The role of government affairs

Frontier AI is making government affairs a more strategic business function.

Companies cannot afford to wait until the rules are final. They need to monitor developments in each important market, engage with policymakers early, and build relationships with regulators.

Government affairs teams also have an important internal role. They must explain how policy changes could affect the business and turn complex developments into practical guidance. They can support consultations while rules are still being developed and help companies plan for possible restrictions.

Frontier AI policy is being developed quickly, often with input from businesses willing to take part. Companies that do not engage may find that important rules have been written without their perspective.

The consequences may go beyond an unexpected compliance requirement. A company could lose access to technology that has become central to its operations.

The question of who can access frontier AI is not going away. Businesses that treat it as a government affairs priority, rather than only a legal or technology issue, will be better prepared for what comes next.

Sources

• Five Eyes cyber security agencies’ joint statement, Australian Cyber Security Centre
• “AI could breach government and business defenses in months, US and its intelligence partners warn,” CNN
• “A Kill Switch for Frontier AI,” Lawfare
• “US AI Export Controls Cause Furor,” CEPA
• “US government order forces commercial suspension of two frontier AI models,” IAPP
• “What Export Controls on Anthropic’s Most Advanced Models Mean for Europe,” AI Frontiers
• “When the Government Pulls the Plug: Anthropic, Export Controls, and the Future of AI Governance,” Volkov Law
• MAS Guidelines on Artificial Intelligence Risk Management, Monetary Authority of Singapore
• “Vietnam Takes a Major Step in AI Governance with New Guiding Decree,” Tilleke & Gibbins