The Anthropocene imperative sounds the alarm, Canadian Prime Minister Kearney urges "Don't passively accept"! Excessive reliance on a few large models highlights risks.

Canadian Prime Minister Mark Carney said that the U.S. export ban has cut off all foreign users' access to the latest artificial intelligence models of Anthropic PBC, highlighting the risks of over-relying on a few powerful AI tools. Speaking to reporters during a visit to Ireland, Carney said, "The situation we are currently facing with Mythos and Fable is a consequence of over-relying on certain models." He was referring to the names of the two AI models mentioned above. He said, "No one has done anything wrong in this incident, but if we passively accept, do not learn lessons, do not build and diversify, then we are wrong." This statement is in line with Carney's consistent emphasis on diversity policy themes since taking office as prime minister. This policy is most directly reflected in Canada's huge trade dependence on the United States - under Donald Trump, the U.S. has imposed high tariffs on key goods such as automobiles and steel. At the same time, the U.S. president has also pressured Canada to repeal digital industry taxes affecting Silicon Valley. Nevertheless, Carney said that the governments of Canada and the United States maintain "good information flow" on AI issues, and "they have identified" some risks associated with Anthropic's latest model. Carney has served as governor of the Canadian Central Bank and the Bank of England, as well as a banker at Goldman Sachs. He also compared the systemic interconnections revealed by the 2008 financial crisis within the banking system. He said, "We are facing similar issues regarding model risks," so redundancy and diversity should be sought. Just as the high interconnections between financial institutions in the past led to systemic contagion once a particular institution failed, the over-reliance on a few large AI models may also lead to systemic fragility. AI will be one of the key topics at the G7 summit in France in the coming days. Carney has already discussed this with French President Emmanuel Macron. Several top AI company executives, including Dario Amodei from Anthropic and Sam Altman from OpenAI, are expected to attend the G7 lunch on Wednesday. Carney said, "We need to make progress on AI issues," but he also warned, "The G7 summit will not hang out a 'mission accomplished' banner." This means that building rules for AI governance will be a long and complex process that cannot be achieved overnight.