Artificial intelligence has moved beyond research laboratories and technology conferences. It has entered the highest level of global politics, where governments are debating who should control advanced models, who should receive access to them, and how much regulation the industry requires.
Those tensions became visible at the 2026 G7 summit in Évian-les-Bains, France. Political leaders met with executives from major AI companies to discuss security, infrastructure, international access and the growing influence of artificial intelligence.
The central disagreement was simple but politically explosive: should the most advanced AI systems circulate freely among allied countries, or should governments restrict access when they believe national security is at risk?
The dispute behind the summit
One of the most closely watched issues involved Anthropic and the US government. The Trump administration raised national-security concerns about foreign access to Anthropic’s most advanced models. Anthropic subsequently disabled access to the affected systems while negotiations continued.
President Donald Trump told reporters that discussions with the company were progressing. His comments followed a meeting between G7 leaders and technology executives, including Anthropic CEO Dario Amodei. However, neither the government nor the company disclosed the full details of the negotiations. Reuters reported on the dispute and the questions surrounding international access.
The disagreement exposed a broader problem. American AI companies want to serve international customers, including close US allies. Governments, meanwhile, increasingly view advanced models as dual-use technology that could support scientific research and economic growth but also strengthen cyberattacks, surveillance or biological threats.
A growing divide between allies
European governments generally support innovation but have placed greater emphasis on regulation, accountability and public protections. The United States has focused more heavily on technological leadership, infrastructure development and limiting barriers to commercial growth.
These differences do not mean that the G7 countries disagree about everything. They broadly support cooperation among democratic nations and recognize the need for secure and trustworthy AI. Earlier G7 commitments called for human-centred development, international standards and efforts to prevent fragmented governance. The G7’s Apulia communiqué also emphasized transparency, risk management and resilient semiconductor supply chains.
The clash concerns how those principles should work in practice.
Europe worries that depending too heavily on American models, cloud services and chips could create a strategic vulnerability. US officials fear that unrestricted international access could allow powerful technology to reach adversaries or be used in ways that threaten national security.
AI companies face pressure from both sides. They need access to global markets, enormous computing infrastructure and reliable energy supplies. At the same time, they are being asked to accept export restrictions, safety evaluations and increasing government oversight.
More than a technology debate
The G7 discussion demonstrated that AI is becoming critical infrastructure. Advanced models may influence military planning, healthcare, finance, scientific research, energy systems and government services. Control over these systems could therefore provide both economic and geopolitical power.
Chips, data centres and electricity are now part of the same debate. A country cannot become a leading AI power without access to advanced semiconductors, large amounts of energy, technical expertise and secure digital infrastructure.
This is why the competition is no longer limited to technology companies. Countries are developing national AI strategies, financing data centres and attempting to protect critical supply chains. Governments are also deciding which allies can access their most advanced technology.
What happens next?
The G7 did not resolve every disagreement. Instead, it revealed how difficult international AI cooperation may become.
Governments want innovation without losing control. Technology companies want global growth without facing a different set of rules in every country. European leaders want greater independence while continuing to cooperate with American companies.
The future of AI may consequently be shaped as much by private negotiations and international agreements as by technical breakthroughs.
The biggest AI story is no longer simply which company releases the next powerful model. It is who receives access, who establishes the rules and which governments ultimately hold the power to say no.
