📰 Key Takeaways

During the G7 summit this Wednesday, French President Macron and Indian Prime Minister Modi publicly expressed concerns about the US potentially unilaterally cutting off other nations’ access to US-made top-tier AI models. Macron directly warned Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and President Trump at lunch: if the US can “flip the switch,” it won’t just hurt European users’ economies — it’ll also backfire on AI companies’ own market reputation.

The spark for these comments was the Trump administration’s blockade of Anthropic’s exports of its newly released Mythos 5 and Fable 5 models, citing national security concerns. The trigger was Amazon alerting the White House that some safety guardrails in these two models could be bypassed. However, multiple cybersecurity experts pointed out that the capability issues cited by the government also exist in other models that remain freely accessible — including OpenAI’s products — yet Anthropic’s models remain frozen from export.

Cohere CEO Aidan Gomez said in a statement that this incident confirms their long-held view: if democratic nations keep relying on a handful of big tech companies, it’ll threaten overall resilience. Digital sovereignty is about who controls the foundational tech that will shape economic security and national sovereignty for the next few decades.

G7 leaders also discussed the possibility of establishing a “trusted partner” mechanism, where non-US nations could access Anthropic and OpenAI’s advanced models simply by committing to use them to strengthen defenses against competitors like China. However, the actual scope and feasibility of this mechanism remain unclear, and whether it would provide sufficient guarantees for Paris or Bengaluru startups is yet to be seen.

💬 JudyAI Lab Perspective

At the G7 summit, Macron and Modi directly warned Anthropic and OpenAI CEOs about supply cut-off risks. For the first time, the geopolitical nature of AI models was placed on the highest diplomatic table — this marks the point where the battle over AI infrastructure sovereignty goes public.

Anthropic’s Mythos 5 and Fable 5 were frozen from export by the US government citing security concerns, while other models with similar capabilities aren’t subject to the same restrictions — this inconsistency reveals a harsh reality all AI builders must face: model access isn’t a “service,” it’s a “license” that can be revoked at any time due to political factors. Cohere co-founder’s “digital sovereignty” argument gets to the core issue: when foundational AI tech is concentrated in a handful of private companies, our reliance on their APIs is essentially betting our business resilience on someone else’s policy decisions. Though the G7 proposed a “trusted partner” mechanism, its actual scope remains unclear, and for non-US startups, there’s currently no protection.

Now is the time to assess how dependent your product is on a single model provider — if the API gets cut off, how long can your service survive? This isn’t hypothetical — it’s already happened with Anthropic.

📅 Source Info

🔗 Further Reading