Mark Carney warns US AI restrictions on Anthropic models expose risks of overreliance

Mark Carney Sounds Alarm on US AI Restrictions

Canadian Prime Minister Mark Carney issued a stark warning Sunday about the risks of dependence on US AI providers, citing the recent shutdown of Anthropic's most advanced models as a critical example of systemic vulnerability1

1

. Speaking to reporters in Ireland ahead of the G7 summit in Évian-les-Bains, France, Carney framed US AI restrictions on Anthropic's Fable 5 and Mythos 5 models as a wake-up call for nations building their economies around AI capabilities they do not control2

2

.

AI giant Anthropic said Friday it took its latest artificial intelligence models offline to comply with a directive from the Trump administration to prevent their use by foreign nationals3

3

. The AI export controls mark the U.S. government's most significant step to date to restrict access to the most advanced AI models. Anthropic released Fable widely this week as a limited version of the even more advanced Mythos, which the San Francisco-based company announced on April 7 and described as so "strikingly capable" that it limited use to select customers due to cybersecurity concerns4

4

.

Drawing Parallels to the 2008 Financial Crisis

"The situation we're in collectively right now with Mythos and Fable is something that can happen with overreliance on certain models," Carney said. "Nobody has done anything wrong in the situation. But we will have done something wrong if we just accept this, don't take the lesson, don't build out and diversify"1

1

. The former central banker then drew a direct parallel to the 2008 financial crisis, saying "we have similar things in terms of model risk" and calling for redundancy and diversity in AI infrastructure—the same principles regulators imposed on the banking system after the collapse of Lehman Brothers2

2

.

The analogy carries considerable weight given Carney's credentials. He served as governor of the Bank of Canada during the 2008 crisis and subsequently became the first non-British governor of the Bank of England, where he spent six years strengthening financial system resilience2

2

. When he warns that concentrated AI dependence mirrors the systemic risks that nearly destroyed the global banking system, the comparison draws on direct experience managing financial crises rather than rhetorical convenience.

G7 Summit to Address AI Governance

Carney made the comments ahead of the G7 summit, where artificial intelligence will be one of the major discussions on Monday night5

5

. He said he spent 45 minutes talking with French President Emmanuel Macron about artificial intelligence on Friday night, though he cautioned there "will not be a mission accomplished banner" that comes out of the summit because the issues are complex1

1

. The remark suggests Canada will push for substantive commitments on diversification in AI technology at Évian rather than symbolic declarations.

Anthropic's Dario Amodei, OpenAI's Sam Altman, and Google DeepMind's Demis Hassabis are all slated to attend a working lunch with G7 leaders on Wednesday2

2

. The Anthropic suspension has given Carney a concrete, real-time example to cite at the summit: a US government decision to invoke export controls against a single AI company cut off access to its most capable Anthropic AI models for every user outside the United States, including allies.

Canada's Push for AI Sovereignty

"You'll hear me say this over and over again. It is never a good idea to have one option," Carney emphasized3

3

. Carney linked the U.S. AI curbs to Canada's broader push to diversify trade and technology. More than 70% of Canada's exports go to the U.S., and Carney has set a goal for Canada to double its non-U.S. exports in the next decade as the Trump administration's trade war causes a chill in investment4

4

.

Canada has been positioning itself on this front. On June 4, Carney launched "AI for All," a $2.3 billion national AI strategy that includes sovereign computing infrastructure, a national supercomputer, and plans to raise business AI adoption from 12% to 60% by 20342

2

. The strategy explicitly frames dependence on foreign cloud providers as a vulnerability, addressing the geopolitical implications of overreliance on American AI providers.

Global Response to AI Export Controls

Carney said there is a "good flow of information" between the Canadian and US governments on AI and acknowledged that Washington has identified "some risks" with Anthropic's latest models2

2

. But his emphasis was on the structural lesson about systemic risks rather than the specific dispute between Anthropic and the US Commerce Department. The Mythos model's ability to surpass human cybersecurity experts in finding and exploiting computer vulnerabilities has raised significant cybersecurity concerns1

1

.

The broader pattern is accelerating globally. The EU published a tech sovereignty package earlier this month curbing US cloud dependence, while India has proposed a $5 billion sovereign AI fund in the wake of the Anthropic suspension2

2

. Britain's Cosine is rallying BT, HSBC, and BAE to build a sovereign frontier model. Carney's Ireland visit also produced a bilateral agreement between Canada and Ireland on AI cooperation, tech collaboration, and food security2

2

. Ireland currently holds the presidency of the Council of the European Union, making it a strategic partner for Canada as it seeks to deepen ties with Europe on AI governance, defence, and critical minerals.

For nations and organizations dependent on American AI providers, the implications of these AI export controls are difficult to ignore. The incident demonstrates how quickly access to critical AI infrastructure can be severed by policy decisions, regardless of commercial relationships or strategic partnerships. As discussions continue at the G7 summit and the USMCA agreement between the U.S., Canada, and Mexico comes up for renewal, the question of AI sovereignty will likely remain at the forefront of international technology policy debates5

5

.