G7 leaders discuss a possible arrangement to let select allied “trusted partners” gain access to advanced artificial intelligence models offered by U.S. firms, according to multiple reports citing diplomatic sources. The talks focus on how G7 member states could use cutting-edge AI capabilities while navigating existing limits on deploying non-U.S.-hosted models. Sources say the discussion includes the idea of a pathway that could broaden allied access through a trusted-partner scheme, potentially involving companies such as Anthropic. The proposal is described as being raised in the context of meetings on the sidelines of G7 summit activities at the French lakeside resort of Evian-les-Bains, with U.S. Commerce Secretary Howard Lutnick among those involved. Several accounts link the potential use case to strengthening cybersecurity defenses, including efforts to better protect against threats from rival states such as China. The reports describe the matter as an ongoing discussion rather than a finalized agreement, with details of which partners would qualify and how access would be governed not yet confirmed publicly.