Canadian Prime Minister Mark Carney used a striking moment in the AI industry to drive home a point about national technology policy: the United States had restricted access to Anthropic's AI, and Canada felt the consequences.
According to AP News, Carney said the US restrictions on Anthropic underscored the risks of depending too heavily on American AI providers. Fortune reported that he specifically pointed to the episode as evidence of the danger of relying too much on US-based companies for critical AI infrastructure.
The backdrop to Carney's remarks is significant. According to Yellow.com, Anthropic's Fable 5 model had outperformed OpenAI's GPT-5.5 in benchmarks before a US government order took it offline — meaning Canada and other countries lost access to what had been a leading AI system, with no say in the matter.
The episode crystallizes a debate that has been simmering in governments around the world: as AI becomes embedded in public services, business, and national security, who controls access to the underlying models matters enormously. A policy decision in Washington can instantly cut off capabilities that other countries have come to depend on.
Carney's warning is a signal that AI sovereignty — the ability of a nation to develop or at minimum access AI systems on its own terms — is becoming a live political issue well beyond Silicon Valley.
For anyone who uses AI-powered tools in daily work or government services, this story is a reminder that the infrastructure behind those tools is subject to geopolitical forces entirely outside their control.