UK EdTech Must Own Its AI Sovereignty Now

UK schools cannot afford to rely on foreign AI providers whose access can be cut off at a moment's notice — the US government's clampdown on Anthropic's Fable 5 is a wake-up call for urgent action on data sovereignty in education technology.

Here's why I think this

The recent US government order forcing Anthropic to suspend access to its Fable 5 AI model is not just an isolated incident; it's a glaring example of how UK schools' AI tools remain vulnerable to geopolitical decisions beyond their control. Anthropic, a leading AI company, was compelled to retract Fable 5 because of concerns over "jailbreaking" its safety features, as mandated by US export controls [3][4]. This means the UK's access to cutting-edge AI can be abruptly severed due to foreign policy or regulatory shifts.

In my experience working closely with UK schools, the assumption that AI tools sourced from US tech giants come with guaranteed reliability and security is dangerously naive. Trust in these platforms is fragile and can collapse overnight. Schools handle sensitive pupil data and require stable, compliant infrastructure that respects UK data sovereignty. Relying on US-based AI providers operating under US law means UK schools are exposed to risks that could compromise data privacy, compliance, and continuity of service [1][2].

LESSO AI was built precisely to address this vulnerability — UK infrastructure, UK curriculum, UK compliance — ensuring schools retain control over their data and AI tools without being subject to foreign government intervention.

What most people get wrong about UK data sovereignty in EdTech

Many school leaders and MAT directors believe that if an AI tool is widely used and powerful, it's automatically safe and suitable for UK education. The prevailing wisdom is that the best AI comes from the US, so schools should adopt these tools to stay competitive and innovative.

This mindset overlooks the critical issue of who controls the AI's infrastructure and data flow. The Anthropic Fable 5 incident exposes a blind spot: access to advanced AI is funneled through a handful of US companies, all operating under US jurisdiction. This means the UK has limited influence over the availability and governance of these tools [2]. The assumption that "cloud-based" or "global" AI solutions are neutral and stable is false. They are geopolitical assets subject to export controls, government orders, and shifting regulations.

Furthermore, many schools underestimate the complexity of compliance with UK data protection laws when using foreign AI services. The nuances of the Data Use and Access Act 2025 and updated KCSIE 2025 guidance mean schools must scrutinize where data is stored, who can access it, and under what legal frameworks. Ignoring this risks breaches and loss of trust [6][7].

What I've seen actually happen in schools

In my work with multiple UK schools and MATs, I've witnessed firsthand the anxiety and disruption caused when AI tools become suddenly unavailable or restricted. One school had integrated a US-based AI writing assistant into its curriculum, only to find the service throttled due to new export restrictions. Teachers lost access mid-term, forcing a scramble to find alternatives and disrupting lesson plans.

Another MAT director shared concerns about pupil data being processed outside the UK, with no clear visibility or control. This lack of sovereignty led them to pause AI adoption entirely until they found a provider committed to UK infrastructure and compliance.

These experiences are not isolated. According to a recent Ada Lovelace Institute survey, 56% of UK respondents want the government to prioritise UK companies for AI, even if it means using less powerful tools [5]. This reflects a growing awareness among educators and policymakers that sovereignty is paramount.

"Relying on foreign AI providers means UK schools are at the mercy of geopolitical decisions that can disrupt education and compromise data privacy."

What I'd do differently

UK education leaders must stop treating AI as a commodity and start treating it as critical infrastructure. This means:

• Prioritising AI platforms built on UK infrastructure with full compliance to UK data protection laws.
• Demanding transparency from vendors about data storage, access, and legal jurisdiction.
• Investing in UK-based AI innovation that understands the unique needs of UK schools and curriculum.
• Developing clear, ratified AI policies by Q3 2026 that address data sovereignty explicitly [StatCallout stat="Q3 2026" label="DfE deadline for ratified AI policies" source="DfE Generative AI Guidance, June 2025" />].
• Collaborating across MATs and local authorities to share best practices and build collective resilience.

We cannot wait for another crisis like the Anthropic shutdown to force reactive measures. Sovereignty must be embedded in procurement and policy decisions now.

The uncomfortable truth

Many in UK education still prefer to chase the latest, most powerful AI tools from US tech giants, blinded by hype and ignoring sovereignty risks. The truth is, this approach leaves schools vulnerable to sudden service withdrawals, data exposure, and compliance failures.

It's uncomfortable to admit that the UK currently lacks sufficient homegrown AI infrastructure for education at scale. But clinging to foreign solutions because they are "better" or "more advanced" is a strategic mistake. Sovereignty is not about settling for less; it's about controlling what matters most — our pupils' data, our schools' autonomy, and our ability to teach without external interference.

If UK education leaders do not demand sovereignty now, they will continue to be powerless when the next geopolitical lever is pulled.

I'm keen to hear how other school leaders and MAT directors are navigating this challenge. Have you faced disruptions from AI providers or concerns about data sovereignty? Let's start a conversation — share your experiences and thoughts on how UK education can take back control of its AI future.

Generate Your School's AI Policy — Free

Mrs J drafts a UK-compliant AI policy for your school in under a minute. Mapped to DfE guidance, DUAA 2025, and KCSIE 2025. Sent directly to your inbox.