What happened
Carney, speaking publicly on Saturday in remarks carried by Crypto Briefing, said the US decision to restrict exports of Anthropic's AI models concentrates dangerous leverage in one jurisdiction. He argued that allied governments have spent two years treating AI access as a commercial question when it is, in his words, a strategic one. The ban itself, announced by Washington earlier this quarter, blocks the cross-border distribution of Anthropic's most capable Claude-tier models to a defined list of foreign entities.
Carney's intervention is the first from a former G7 central bank chief framing the move as a sovereignty problem rather than a competitiveness one. He did not call for retaliation. He did call for diversification.
Why it matters
Until now, US AI export controls have focused on hardware. Nvidia H100s. Wafer fab equipment.
The Anthropic ban is the first serious extension into the model layer itself, and that changes the calculus for anyone building on top of US-hosted frontier AI. Carney's framing matters because it gives allied policymakers political cover to fund alternatives. Read it as a signal that Ottawa, Brussels, and London will push harder on domestic compute and on open-weight model strategies.
For crypto, the read is direct. Decentralized inference networks, distributed training protocols, and tokens that promise censorship-resistant AI compute have spent two years pitching exactly this scenario. The pitch just got a real-world data point from someone with central bank credibility.