AI at G7: 'Trusted partners' plan being discussed amid US export ban on Anthropic's AI models

Artificial Intelligence is at the centre of conversations at the G7 Summit in Évian-les-Bains, France, with member countries discussing access to cutting-edge models. There is a scramble for new international "deals" and frameworks as G7 leaders reportedly talked about a “trusted partners” plan where selective access to advanced AI models from US firms such as Anthropic is granted. The matter assumed high priority after Anthropic disabled access for all users to Fable 5 and Mythos 5 on Friday. This was done after US President Donald Trump ordered Anthropic to block foreign nationals from accessing its most advanced AI models. He said that it had to be done due to cybersecurity and national security concerns.

Financial Times reported that a US delegation led by Commerce Secretary Howard Lutnick and European diplomats is holding discussions to have a system where close, trusted G7 allies, either companies or countries, would be able to access American AI models that are currently not open to export. Such an agreement would allow G7 countries to use the AI technology to strengthen cybersecurity defences against rivals such as China.

Anthropic's Fable 5 and Mythos 5 models

Fable 5 and Mythos 5 are yet to be released models by Anthropic. Cybersecurity experts have raised concerns about Mythos, which has been designed to find flaws in computer code. They believe that it may lead to a dramatic escalation in attacks on banks' technology systems. An export control directive on Friday told Anthropic to suspend access to its latest AI models, Fable 5 and Mythos 5, "by any foreign national, whether inside or outside the United States." Both models were released just three days before the order was issued.

Also Read: AI, data centres and digital colonialism: How intelligence became a commodity

Countries are now seeking access to Mythos to study how the model could impact safety and find kinks in the computer systems. Before Trump's order, select organisations in "more than 15 countries" were granted access to Mythos so they could use it to detect vulnerabilities in their computer systems.

Also Read: Company accidentally blows away Rs 4,800 crore on Claude AI in a month because of a tiny oversight

Technology independence debate at G7

The export restrictions have led to debate about freedom of technology and caused noticeable friction among G7 members. Henna Virkkunen, the European Commission’s tech chief, has called for removing the "discriminatory" tech measures. However, it has come as a wake-up call for European leaders who are pushing for quick funding of the domestic European AI alternatives. Earlier, Canadian Prime Minister Mark Carney said that the export restrictions highlight the dangers of overreliance on a limited number of American providers.

Also Read: WATCH | Modi meets Trump on G7 sidelines in first in-person interaction since 2025

AI leaders meet for formal working lunch

Notably, AI executives from Anthropic, OpenAI, and Google, are in France and scheduled to attend a working lunch on Wednesday to address issues like regulation, AI infrastructure, and networks. In an unprecedented move for a G7 summit, the world’s top AI executives are gathering for sessions. The main stage working lunch will include OpenAI CEO Sam Altman, Google DeepMind's Demis Hassabis, Anthropic’s Dario Amodei, and Mistral AI’s Arthur Mensch, alongside executives from Meta, Salesforce, and Cohere. They will be discussing pointers on AI infrastructure, especially its energy demands and the sustainability of the data centres.

AI safety at G7

Two other points on the table include protecting minors and trust. French President Emmanuel Macron has led an agenda item of a joint declaration aimed at mitigating the risks of AI-generated harmful content for minors. Other member nations, including Indian Prime Minister Narendra Modi, are calling for "humanity-first" transparency and trust frameworks, so AI deployment closes economic productivity gaps and does not widen them.