Concerns over the United States' potential to restrict access to advanced artificial intelligence models have been voiced by global leaders at the recent G7 summit. French President Emmanuel Macron and Indian Prime Minister Narendra Modi were among those who highlighted the risks of reliance on American-developed AI, following a recent decision by the US government to block the export of new models from the AI firm Anthropic.
President Macron, speaking to G7 leaders and prominent AI executives, including OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei, warned of the significant economic repercussions should the US be able to 'turn off the switch' on AI access overnight. Such a move, he argued, would not only damage European economies dependent on these technologies but could also harm the American AI companies themselves by eroding international trust and demand. This sentiment was echoed by Prime Minister Modi, who, according to reports, expressed apprehension about the Anthropic restriction and emphasised the importance of democratic nations retaining unrestricted access to top AI models for the protection of critical national infrastructure.
The specific incident triggering these anxieties involves the Trump administration's decision to prohibit Anthropic from exporting its latest Mythos 5 and Fable 5 models. This block was reportedly implemented on national security grounds after Amazon raised concerns with the White House regarding potential bypasses of certain safety guardrails. While some cybersecurity experts have pointed out that similar capabilities exist in other readily available models, including those from OpenAI, Anthropic's models remain inaccessible internationally. This episode has starkly exposed the vulnerability faced by international companies and governments building upon US AI infrastructure, highlighting the possibility of access being revoked suddenly and without detailed explanation.
The broader implication, as articulated by Aidan Gomez, co-founder and CEO of Canadian enterprise AI firm Cohere, is that continued dependence on a limited number of major US tech companies poses a threat to national resilience and digital sovereignty. Gomez stated that 'digital sovereignty is not just about market competition or any one company or nation. It’s about who controls the foundational technology that will shape our economic security and national sovereignty for decades to come.'
In response to these concerns, G7 leaders also discussed the potential establishment of a 'trusted partners' scheme. This initiative aims to grant non-US nations access to advanced AI models from firms like Anthropic and OpenAI, effectively creating an open trade network designed to circumvent unilateral US restrictions. The proposed scheme would involve both countries and companies as trusted partners, provided they leverage these models to bolster defences against rival powers, such as China. However, the practical scope and effectiveness of such a scheme, particularly for smaller European or Indian startups whose products could be disrupted without warning, remain unclear.
Despite these discussions, Macron suggested that Washington would benefit from supporting such a scheme and ensuring broader access to models like Mythos. The underlying message is clear: the long-term viability of US AI exports hinges on a predictable and reliable access policy, as international customers are unlikely to invest in technology that could be withdrawn at a moment's notice. This comes as non-US countries, including those in Europe, strive for greater AI sovereignty, a goal made increasingly challenging by the rapid advancements of American models.