Carney Says Anthropic Ban Shows AI Dependence Risks

Canadian Prime Minister Mark Carney says the U.S.-driven shutdown of Anthropic’s most advanced AI models is a warning to allies that depending too heavily on a small number of American technology providers can create serious strategic risks. The remarks come just as artificial intelligence rises on the agenda at the G7 summit in France.

What happened

Anthropic said it abruptly disabled access to its newest high-end models, Fable 5 and Mythos 5, after receiving a U.S. export control directive that barred access by foreign nationals. Because the rule reportedly applied both outside the United States and to foreign nationals inside the country, the company said it shut the models off broadly to ensure compliance.

Other Anthropic services were not reported as part of the same shutdown. The move immediately turned a model-access issue into a wider debate about sovereignty, resilience, and who ultimately controls critical AI infrastructure.

What Carney said

Speaking in Ireland before traveling to the G7 summit, Carney said the disruption should serve as a cautionary example for countries that rely too much on a narrow group of AI suppliers. His message was straightforward: allies need more than one option, and they need to diversify.

That framing is significant because it treats advanced AI models like strategic infrastructure. In practical terms, it suggests governments may increasingly view model access the same way they view cloud capacity, semiconductors, or energy security.

Why Canada is pushing AI sovereignty

Carney’s comments align closely with Canada’s recently announced “AI for All” strategy. The plan aims to expand AI adoption, grow the domestic AI economy, and reinforce Canadian sovereignty through investments in compute, cloud, talent, and public-interest infrastructure.

The Canadian government says the strategy targets an additional $200 billion in economic growth and 250,000 AI-related jobs over five years. It also includes plans for sovereign compute capacity and stronger partnerships with trusted allies, showing that Ottawa wants AI capability to be both an economic asset and a national resilience issue.

Why this story matters beyond Canada

This incident highlights a growing tension in the AI race: the most advanced models may be globally desired, but access to them can still be shaped by national security rules in the country where they are built. For companies, universities, and governments outside the United States, that creates a new layer of uncertainty when choosing providers or building services on top of frontier models.

If more export controls follow, countries may accelerate local compute projects, back domestic AI firms, or spread workloads across multiple vendors. That could reshape the global AI market from one centered mainly on capability to one increasingly shaped by jurisdiction, compliance, and strategic trust.