- Allies Explore Possible Framework for Wider AI Access
EVIAN-LES-BAINS, France: Leaders of the Group of Seven (G7) nations have discussed a proposal to allow selected “trusted partners” to access advanced artificial intelligence models developed by major US technology companies, including Anthropic, according to three diplomatic sources.
The discussions, held on the sidelines of the G7 summit in the French lakeside resort of Evian-les-Bains, focused on creating a possible framework that could expand access to powerful AI systems while addressing Washington’s national security concerns.
Talks Held With US Officials Over AI Restrictions
Several representatives from G7 countries raised the idea of broader access to advanced AI models during meetings with US officials, including Commerce Secretary Howard Lutnick, during the opening summit dinner on Monday, one diplomatic source said.
A second source said the proposed “trusted partners” could include either governments or companies, though details remain under discussion.
The move comes after US President Donald Trump directed Anthropic to restrict access to its most advanced AI models for foreign nationals, citing concerns over national security.
Following the order, Anthropic disabled access for all users to its latest models, including Fable 5 and Mythos 5.
AI Access Seen as Key to Cybersecurity Cooperation
A potential agreement could allow G7 nations to use advanced AI tools to strengthen cybersecurity defenses against strategic rivals, including China.
A White House official said the administration remains in close communication with allies and is focused on addressing national security issues linked to Anthropic’s technology.
AI companies, including Anthropic, OpenAI and Google, are expected to join G7 discussions later in the week to address issues such as AI regulation, infrastructure development and digital networks.
Concerns Over Advanced AI Capabilities
Cybersecurity experts have raised concerns about the capabilities of Anthropic’s Mythos model, which is designed to identify weaknesses in computer code. Analysts warn that such technology could potentially be misused to accelerate cyberattacks against financial institutions and critical infrastructure.
The European Union has also expressed interest in gaining access to the model to evaluate its risks and implications.
Before the restrictions were introduced, Anthropic had provided access to Mythos to selected organizations in more than 15 countries. These included institutions in healthcare, telecommunications, energy and water sectors, where the model was used to identify potential vulnerabilities in computer systems.
The proposal for a “trusted partners” access system was first reported by the Financial Times.
BY: The Times Union
