Key Takeaways
- French President Emmanuel Macron and Indian Prime Minister Narendra Modi voiced concerns at the G7 summit about potential US control over access to advanced AI models.
- These concerns were heightened by a recent incident where American AI company Anthropic temporarily restricted access to its advanced models, Fable 5 and Mythos 5, for foreign nationals due to a US government order citing national security.
- The incident underscores a growing international push for "AI sovereignty," with nations aiming to reduce dependence on a few dominant AI providers and develop their own indigenous capabilities and regulatory frameworks.
- Leaders at the G7 summit, including AI executives from major companies, discussed the need for global standards, trusted partnerships, and secure AI infrastructure to balance innovation with national control.
The world's most powerful artificial intelligence tools are largely developed and controlled by a handful of American companies. While many nations eagerly seek to leverage these cutting-edge advancements, a significant and growing apprehension shadows this reliance: the fear that access could be revoked overnight. This concern, brought sharply into focus by French President Emmanuel Macron and Indian Prime Minister Narendra Modi at the recent G7 summit, gained stark realism following a disruptive "blackout" by US-based AI firm Anthropic.
The G7 Summit: A Call for AI Sovereignty
At the 52nd G7 Summit in Évian-les-Bains, France, AI emerged as a critical topic alongside global security and economic challenges. During discussions, both President Macron and PM Modi highlighted the geopolitical vulnerabilities stemming from concentrated AI power. Their message was clear: while American AI is desired, the potential for the U.S. to unilaterally "turn off the tap" on these essential technologies is a major concern for national security and economic stability.
PM Modi, attending the summit as India joined as a partner country, emphasized the need for inclusive, open, and globally accessible AI development. He warned against the misuse of AI and called for strengthening global cooperation to counter issues like deepfakes, misinformation, and cyber fraud, stressing that without safeguards, children could be exposed to risks. Macron echoed these sentiments, stating that AI frontier models must not fall into the hands of authoritarian regimes and that democracies must cooperate on this topic.
The leaders' anxieties stem from a world where the digital economy's commanding heights—from advanced semiconductors to frontier AI models and cloud infrastructure—remain anchored within a few jurisdictions. This centralisation, while offering global access to technology, also creates structural points of dependency and vulnerability.
Anthropic's Blackout: A Concrete Warning
The urgency of these discussions at the G7 was amplified by a recent incident involving Anthropic, a prominent American AI safety research company and developer of the Claude AI models. On June 13, 2026, Anthropic disabled access to its most advanced artificial intelligence models, including Fable 5 and Mythos 5, for all foreign nationals. This unprecedented move followed an order from the US government, citing national security concerns, to prevent the technology from reaching foreign hands, "whether inside or outside the United States."
The company stated that the government issued the order after discovering a potential "jailbreak" or bypass in Fable 5, a recently released version of Mythos, which had been blocked from performing cybersecurity tasks. Anthropic expressed disagreement, arguing that such a strict standard, if applied broadly, could halt future model deployments across the industry. This incident, seen by many as a "real-world example" of the fears articulated by Macron and Modi, demonstrated how a policy decision in Washington could instantly cut off access to critical AI infrastructure for users worldwide.
It's important to note that this specific "blackout" was distinct from earlier service disruptions that Anthropic's Claude AI experienced in the preceding weeks and months, including elevated error rates and outages impacting various models. For instance, on June 5, 2026, Claude experienced a roughly two-hour outage, and a significant worldwide outage also occurred in March 2026. While those were technical issues, the June 13th restriction was a direct governmental intervention, highlighting the geopolitical dimension of AI control.
The Rising Tide of AI Sovereignty
The concerns voiced at the G7 and underlined by the Anthropic incident have intensified the global push for what is known as "AI sovereignty." This concept refers to a nation's ability to make independent decisions about the deployment, use, and adoption of critical AI infrastructure, rather than being entirely reliant on foreign providers.
Motivations for AI sovereignty are diverse, including protecting national security, ensuring economic competitiveness, retaining talent, and guaranteeing that AI systems reflect local cultural, linguistic, and social values. Countries are increasingly wary of "vendor lock-in" and the potential for their digital future to be dictated by external powers.
European Union's Approach
The European Union has been a frontrunner in establishing a comprehensive framework for digital and AI sovereignty. Building on initiatives like the GDPR, the EU AI Act, which classifies AI systems by risk and imposes conformity obligations, aims to set global standards for trustworthy AI.
More recently, the EU introduced the Cloud and AI Development Act (CADA) as part of its "European technological sovereignty package." CADA defines cloud and AI sovereignty across four assurance levels, primarily for public sector bodies. It evaluates providers based on infrastructure location (data processed and stored in the EU), independence from third countries, transparency over software supply chains, and even EU ownership and control. This framework aims to strengthen Europe's competitiveness in the cloud and AI ecosystem and reduce strategic dependencies.
India's National AI Strategy
India, recognizing AI's transformative potential, has also been developing a robust national AI strategy. Launched in 2018, the "National Strategy for Artificial Intelligence" focuses on leveraging AI for inclusive growth, developing the research ecosystem, and promoting adoption across sectors like agriculture, health, and education.
India's approach balances fostering innovation with mitigating risks, emphasizing ethical AI development, data privacy, and transparency. The country aims to build indigenous AI capabilities, democratize access to computational resources, and foster local AI talent. The "India AI Mission," approved in March 2024, further reinforces these goals, supporting startups, industry collaboration, and socially impactful AI projects.
France's Drive for Autonomy
Beyond its calls at the G7, France has its own national strategy for AI, adopted in November 2021 as part of the "France 2030" plan. This strategy aims to expand the pool of AI-trained talent, accelerate R&D, and integrate AI into the economy, focusing on areas like embedded AI and trustworthy AI. France plans significant investments—over €109 billion for AI infrastructure projects—and is collaborating with companies like NVIDIA and Mistral AI to build cutting-edge compute platforms.
A recent development saw France's domestic intelligence service announcing it would replace AI data tools from the US company Palantir with those from a domestic provider, ChapsVision. French Prime Minister Sébastien Lecornu explicitly stated this move was to avoid "strategic dependencies," emphasizing that France must "build real autonomy" and not "depend on the goodwill of certain partners, who are capable of turning off the access tap" for artificial intelligence.
The Path Forward: Balancing Innovation and Control
The discussions at the G7 and the responses to incidents like Anthropic's access restrictions highlight a critical challenge in the age of AI: how to balance rapid innovation with national control and security. While a completely "self-sufficient" AI ecosystem is difficult to achieve given the global nature of supply chains, talent, and data, nations are seeking strategic autonomy.
This involves managing AI dependencies layer by layer, from compute capacity and data access to foundational models and governance. Leaders at the G7 summit discussed plans to grant select "trusted partners" access to advanced US AI models, potentially creating a pathway around blanket restrictions and allowing for stronger cybersecurity defenses. AI executives from Anthropic, OpenAI, and Google also attended the summit, engaging in discussions about regulation, AI infrastructure, and global standards.
The ongoing dialogue points towards a future where international cooperation, clear global standards, and diversified AI ecosystems will be crucial. The goal is not isolation, but rather the ability for nations to act deliberately within an interdependent system, ensuring that the transformative power of AI serves global progress without compromising national sovereignty.
Frequently Asked Questions
What is AI sovereignty?
AI sovereignty refers to a nation's capacity to make independent decisions regarding the development, deployment, and use of critical artificial intelligence infrastructure and technologies, reducing reliance on foreign providers.
Why are world leaders concerned about American AI access?
World leaders are concerned because many advanced AI models and infrastructure are developed and controlled by US companies. They fear that the US government could, for national security or other reasons, restrict access to these vital technologies, impacting their nations' economies, security, and digital autonomy.
What was the Anthropic blackout, and how did it relate to these concerns?
The "Anthropic blackout" refers to a specific incident on June 13, 2026, where American AI company Anthropic disabled access to its advanced AI models (Fable 5 and Mythos 5) for all foreign nationals. This was in response to a US government order citing national security concerns. The incident provided a tangible example of how access to critical AI could be cut off, validating the fears expressed by leaders like Macron and Modi.
How are countries like the EU, India, and France addressing AI sovereignty?
The EU is implementing the AI Act and the Cloud and AI Development Act (CADA) to regulate AI and strengthen its own cloud and AI infrastructure, emphasizing data localization and supply chain control. India has launched a National Strategy for AI and the India AI Mission to foster indigenous AI capabilities, ethical development, and inclusive growth. France is investing heavily in its national AI strategy, developing domestic talent and infrastructure, and actively replacing foreign AI tools with local alternatives to reduce strategic dependencies.
