Trusted, Sovereign AI: Why nations can no longer afford to depend on others for their intelligence

From Dependency to Capability

When AI systems are designed abroad and hosted in foreign clouds, nations inherit risks they cannot easily control: opacity in algorithms, exposure of sensitive data, and vulnerabilities to policy shifts or export restrictions by the owning state. In defence, aviation, energy and public services, the domains where data sensitivity and operational continuity are paramount, this is not a theoretical risk. It is a strategic liability. 

AI Sovereignty replaces import-led dependency with verifiable trust. It means exercising control over the entire AI lifecycle under domestic law, mission policy and societal values. It is, in effect, the AI equivalent of energy security: nations must decide what remains domestic, what is shared, and how collaboration occurs. 

A Framework Built on Four Pillars

Thales' approach to AI Sovereignty is organised around four foundational pillars that translate this strategic ambition into engineering practice: 

• Trust: Every AI decision must be explainable, auditable and defensible. Through the TrUE AI (Trusted, Understandable and Explainable AI) framework embedded in CortAIx, Thales ensures that systems remain transparent, secure and under human oversight from design to deployment.
• Sovereignty: Nations retain end-to-end control over data, models, pipelines and infrastructure. This includes jurisdictional assurance so that sensitive data stays on national assets, alongside federated learning environments that prevent data from ever leaving national boundaries.
• Innovation: Sovereignty is not isolation. Open collaboration with startups, academia and international partners can thrive within clear governance structures. The focus is to localise engineering, transfer know-how, and keep critical intellectual property within national ecosystems.
• Future Readiness: AI architectures must absorb change without surrendering control. Modular, policy-aware designs, frugal inference methods, and investments in post-quantum cryptography ensure that today's sovereign systems remain resilient as the technological landscape evolves. 

The Middle East: A Region That Cannot Wait

For the Middle East, AI Sovereignty is not a distant aspiration, it is an accelerating strategic reality. AI strategies embedded into national visions signal a shared commitment to build domestic capability, develop indigenous talent, and govern technology on national terms. 

The region is rapidly scaling smart government services, advanced defence ecosystems, aviation hubs and energy transitions — precisely the domains where data sensitivity and regulatory alignment demand sovereign control. With large-scale investments in talent, R&D and national compute capacity already under way, the governance framework for AI Sovereignty is the missing piece that ensures this investment compounds domestically rather than enriching foreign platforms. 

Thales' presence in the region reflects this ambition. As the Group looks to extend its CortAIx ecosystem to the Middle East, the initiative will serve as a bridge between global research excellence and national co-development, strengthening local talent pipelines and ensuring that the region benefits directly from Thales' trusted, mission-ready AI frameworks. 

Thales is not simply adopting AI, it is building the sovereign capacity to govern it.

CortAIx: Operationalising Trusted AI

At the heart of Thales' strategy is CortAIx, the Group's global AI accelerator and innovation engine for mission-critical environments. Uniting over 800 engineers and PhD researchers across Canada, France, the UK, and Singapore, CortAIx brings together world-leading expertise in research, sensors and systems integration to deliver AI that is trusted, frugal and resilient under constraint. 

Thales leads the European ranking of AI patent filers, and its capabilities are embedded in more than 100 operational solutions, spanning electronic warfare, airborne target recognition, maritime surveillance, air traffic management and cyber operations. What distinguishes Thales is its 'magic square': the balance of performance, frugality, security and ethics that positions it uniquely among global technology players. 

Frugal AI is a defining differentiator. Some airborne sensors generate over a terabyte of data per second, yet only a fraction can be stored or transmitted. Through model quantisation, pruning, distillation and edge-first inference, Thales delivers AI that operates in real time with sub-100-millisecond responses under size, weight and power constraints, without the energy demands typical of large commercial models. 

Talent: The Human Foundation of Sovereignty

Compute and data can be procured. Trusted expertise must be developed nationally. Thales recognises that human capital is the limiting factor in achieving sustainable AI sovereignty, which is why the Group invests in graduate pathways, upskilling programmes and joint research clusters across the regions where it operates. 

In the UAE, Thales has collaborated with Sorbonne University Abu Dhabi on hackathons focused on AI awareness and capability-building — an example of how academia-industry partnerships can inspire local innovation while reinforcing national strategy. The Group's internal targets reflect the same philosophy: up to 30% national AI specialists within UAE teams and 60% local talent in Saudi Arabia, forming the enduring human foundation of technological independence. 

Sovereignty in Action: Real-World Impact

AI Sovereignty is not abstract. The following capabilities illustrate how the four pillars translate into measurable operational advantage: 

• GoldenAI (Electronic Warfare): Automatically classifies 90% of intercepted radar emissions so that operators can focus on the ambiguous 10%, delivering up to 10× efficiency gains while keeping sensitive EW signatures within sovereign infrastructure.
• Talios Pod (Airborne Targeting): Embedded AI analyses real-time imagery and identifies targets up to 100× faster, reducing crew workload and preserving human-in-the-loop control on national combat aircraft.
• GenAI4SOC (Cyber Operations): Generative AI surfaces correlated anomalies and produces explainable incident reports in real time, ensuring that critical security operations centres retain full control over threat intelligence pipelines.
• TopSky Sequencer (Air Traffic Management): AI optimises flight sequencing in real time, cutting system configuration time by 10× and reducing CO₂ emissions, demonstrating how AI Sovereignty supports both economic and sustainability goals.
• MINDS (Activity-Based Intelligence): AI-driven detection and generative reporting transform raw imagery into actionable intelligence, delivering 4× faster analysis and 70% reduction in reporting time in compliance with NATO standards. 

An Agenda for Action

For governments and industry leaders seeking to convert AI ambition into sovereign capability, Thales identifies five practical priorities: 

• Establish national AI assurance frameworks aligned to standards such as ISO/IEC 42001, ensuring explainability, auditability and continuous testing.
• Operationalise end-to-end data sovereignty with jurisdictional control, access governance and federated training environments.
• Adopt dual sovereignty: proprietary models for sensitive workloads, vetted open-source stacks for non-critical domains.
• Prioritise frugal AI to guarantee continuity under constraint while cutting energy use.
• Invest in national talent pipelines through early-career programmes, retraining initiatives and specialist academies. 

AI Sovereignty is already delivering measurable impact, faster detection and analysis, lower cognitive load, greater survivability and reduced energy footprint. The path forward is clear: autonomy where it matters, collaboration where it benefits. With CortAIx, Thales stands ready to help nations turn that vision into enduring national capability.