Locked in: A realist's agenda for digital sovereignty

The US has previously restricted access to the most advanced artificial intelligence (AI) and asked labs to hold back new models from the UK AI Security Institute. It is now a realistic possibility that critical systems get cut off, that economic gains from transformative technology flow overseas, and that the UK is unable to regulate an increasingly dangerous technology on its own terms. 

In response, governments worldwide are investing billions in ‘sovereignty’, defined broadly as the ability for a country to control and shape the technology it relies on. Yet there remains limited clarity around what level of control is achievable for the UK. 

One camp argues the UK cannot compete and should accept dependence, negotiating the best access terms it can get. The other camp argues the UK can aim for interdependence and with it more control, spreading reliance beyond just the US and building a UK technology stack that others depend on. 

Even with ambitious reform, the UK is unlikely to be in control of a chokepoint on its own. The best route is to manage dependence on US frontier models by working with other middle powers who bring substantive pooled capability. Together, these countries can become indispensable.