G7 leaders discuss 'trusted partners' access to cutting-edge US AI models, sources say

G7 leaders have discussed a plan to grant select trusted partners access to advanced US AI models from companies like Anthropic, according to diplomatic sources familiar with the matter. The discussions took place during the recent G7 summit, though no formal agreement has been reached.

The plan is seen as a way to maintain strategic advantage in AI while managing risks of misuse. Anthropic, an AI safety company, was named as one of the firms whose models could be shared under the framework.

The G7 has previously issued statements on AI safety and governance, but this marks a concrete step toward controlling access to frontier capabilities. The proposal faces potential hurdles over intellectual property and equitable access among allies.