On June 17, the Financial Times reported, citing informed sources, that Amodei made his remarks during a G7 leaders' luncheon in Évian-les-Bains, France, stating, "We must overcome the temptation to divide." He acknowledged the efforts of various countries to prevent AI from falling into the wrong hands but emphasized the need for democratic nations to work together.
Also present at the meeting were Sam Altman, CEO of OpenAI, and Demis Hassabis, CEO of Google DeepMind. Altman stressed that all participating countries should be provided with cyber defense tools.
Amodei, Altman, and Hassabis all called for U.S.-led cooperation in AI model development. They warned that if democratic allies were to fracture, it could increase risks related to biological terrorism and cybersecurity.
Last week, the U.S. blocked the export of Anthropic's Mythos and Fable models, citing national security concerns. This decision followed reports suggesting that the models could bypass safety mechanisms.
The move has raised fears in Europe and Silicon Valley that the U.S. government could prevent non-American clients of allied nations from accessing the latest AI models.
French President Emmanuel Macron remarked that the Anthropic situation clearly illustrates the issues facing the U.S. and its G7 allies. He warned that if the U.S. could suddenly turn off access, it could also harm American big tech companies leading the AI race.
Macron noted that discussions with G7 leaders, including President Trump, were productive, but no conclusions had been reached on how to handle advanced AI models. He stated that stronger regulations on AI are necessary and highlighted the risks of non-cooperation among democratic nations. He expressed intentions to establish a discussion and cooperation platform among several democratic countries in the coming months to address the risks posed by AI and to set common standards.
Indian Prime Minister Narendra Modi also expressed concern over the U.S. blocking access to Anthropic's AI models, emphasizing that democratic nations should have access to AI models to protect critical infrastructure.
The meeting also addressed how countries can secure AI technology supply chains that they can control. Arthur Mensh, CEO of French AI company Mistral, raised the issue of whether one can be assured that they won't be blocked by others when supply chains are interlinked, noting that this concern was primarily voiced by attendees from countries other than the U.S.
The Financial Times highlighted that this discussion has elevated the issue of U.S. AI export controls and allied nations' access to technology to a major agenda item at the G7 level. It noted the clash between the security rationale for controlling the spread of advanced AI models and the industrial and diplomatic need to maintain technological cooperation among allies.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.