AI Thinking

That ‘sovereignty’ issue you thought was super low risk

Katie Lips, CEO and Founder of Rebel Futures/16 June 2026
That ‘sovereignty’ issue you thought was super low risk

When assessing vendors for any digital programme or platform, you run through a familiar checklist: price, feature evolution, interoperability, and lock-in. You assess the risk of price increases, obsolescence, a vendor being acquired, security breaches and so on.

Government intervention and sovereignty risk are usually on that matrix too, but somewhere way down near the bottom, and classed as ‘low’ or ‘very low’ risk. They’re there for completeness, but not because you actually think they could affect your strategy.

On June 12th, the US government ordered Anthropic to suspend all access to two of its most capable models (Fable 5 and Mythos 5) for everyone who’s not a US national. Of course, proving nationality at the point of use is impossible, so it suspended access for every customer, everywhere, within hours.

That super low-risk item on your risk register just got bumped to the top. And it became urgent.

The US Government’s intervention demonstrates that we need to reconsider risk in our vendor assessment and AI platform and architecture strategies.

Whatever you think of the model capabilities, of Anthropic’s safety record, of their ‘uber powerful model marketing strategy’, the risk to your AI in your day-to-day operations, to the AI that powers your product, is real. Most organisations have never modelled nationality as a variable in their AI architecture at all.

Technical leaders should treat single-provider dependencies as a risk. If your product or operational AI tooling is built on one frontier model with no abstraction layer beneath it, you don’t have much of a response to an event like this. Instead:

  • Consider multi-model strategies - Make sure you have a tested failover path.
  • Prioritise abstraction layers - Build a model-agnostic interface and buy yourself options, even if you’re just using one provider in practice.
  • Add the scenario to your incident response plan - Create a plan for “our model provider is removed from service by government without warning”.

C-suite leaders need to think differently, too. Sovereignty risk in AI vendor relationships deserves an owner and a review trigger.

  • Ask vendors new questions - not simply “is this safe and reliable” but “what is your jurisdictional exposure?” When much AI comes from USA, you might not get much choice, but it at least needs consideration. (This is also a strong argument for investment in frontier AI in other regions, of course.)
  • Review contracts - Force majeure, SLAs and liability clauses written for outages or insolvency may not cleanly cover fast, unilateral, government-mandated suspension.
  • Brief the board - They probably already know, but time spent now costs far less than a board being surprised by this later.

For organisations that take this seriously, we are likely to see three shifts.

Organisations will exercise more caution in their architecture design. I doubt we will see any less AI adoption, but the US intervention will force everybody to scrutinise single-provider dependencies much harder.

More attention will be given to building in redundancy where it matters. It won’t affect everything to the same degree; in operational and productivity tools, where an outage is inconvenient rather than existential, it will be possible to stay single-provider for longer.

To manage risk, organisations will undertake jurisdictional mapping exercises. It’s now more important to have a clear internal picture of where AI dependencies sit, geographically and legally. Very few organisations have done this properly so far.

What’s unlikely is a wholesale retreat from frontier AI. The commercial case for AI in operations, in product, and, as a true differentiator hasn’t changed. It’s just that the “what happens if the thing we thought would always be available, isn’t?” needs a much better answer.

Rebel Futures helps business and technology leaders navigate AI-driven disruption. If this raised a question about your organisation’s AI vendor risk, let’s talk.

Katie Lips is CEO and Founder of Rebel Futures. She helps organisations grasp the opportunities at technology inflection points, from the early web and pre-iPhone mobile to cloud-based SaaS and now AI. She has held senior roles at EY and Aimia, and was Head of Digital Policy at Which?. Katie holds a Harvard Kennedy School qualification in Leading in Artificial Intelligence and is a Founder Institute graduate.

Related: explore our work on AI Strategy.

base44
Edit with Base44