The Ministry of Justice (MoJ) adopting OpenAI tech is a great target indicator for the future of AI in big government.
There’s often a collective intake of breath and a certain level of anxiety when big government and big tech get in bed together, particularly around AI. There are still some very necessary conversations to be had about artificial and generative AI, not least of which is a question on how they’re trained, and how the data to train them is accessed and used.
These questions came to the fore when the government announced Humphrey, an AI model for the civil service trained on models from OpenAI, Anthropic and Google. Detractors said this was dangerous, given the government had no overarching compliance directives with these companies.
Adoption of AI models from our elected officials, while there are still legal cases going on left, right and Brentford, can be seen as a premature green light by some, while others claim they’re late to the party and risk losing an advantage in cost saving, manpower and efficiency by not adopting it sooner.
Let’s put a pin in all of those unsolved anxieties for a second and concentrate on the news regarding the MoJ.
Last week’s announcement that OpenAI will be expanding its British operations and hosting data on UK soil is something worth paying attention to, not just for what it promises, but for what it signals about the future of AI legitimacy, privacy and power.
The deal, announced by Deputy Prime Minister David Lammy, marks a big step in an area outside of just business as usual for the MoJ. For the first time, British businesses will be able to host their OpenAI data on UK-based servers. In government speak, this is about “unlocking productivity” and “boosting the economy.” In human speak, it means our AI conversations, data and workflows won’t automatically end up crossing the Atlantic to be stored in the U.S., as was previously the case.
That’s huge, for a number of reasons.
For years, one of the biggest blockers to adopting AI, particularly in sectors like healthcare, law, education and justice, has been data security. If your conversations with an AI tool leave the UK, they fall under different laws, different privacy protections and different accountability standards. Storing them locally, under British regulation and the ICO’s oversight, makes it far easier to integrate AI responsibly across industries that handle sensitive data.
So where the press release says, “British businesses will be able to host data on secure, sovereign servers,” that’s important. Sovereign data hosting opens up entire areas of business that would have otherwise been excluded from AI on the grounds of compliance. It isn’t a green light for unfettered use of AI by your bank, for example, but it’s moving us closer to AI being business as usual.
The MoJ Goes In Hard
It’s not just business either. The Ministry of Justice is balls deep in this, rolling out AI tools to over a thousand probation officers to automate admin tasks like transcription and case notes, saving what they claim will be “up to 240,000 days of valuable time every year.”
That’s not small. If you’ve ever worked in a compliance-heavy, paperwork-driven workplace, you’ll know the level of box-ticking and note-taking is astronomical. So anything that frees up time for people to actually do their jobs, in this case supervising offenders and reducing reoffending, is worth celebrating.
“New AI tools are already allowing our brilliant probation officers to spend far less time filling out paperwork and far more time face to face with offenders, making them less likely to reoffend.”
What’s being presented as a productivity revolution in a single government department is also a huge shift in how much the UK government is willing to rely on private AI infrastructure. Other areas of government will be watching this as a litmus test to see how they too can save time and money.
Sam Altman, OpenAI’s CEO, said: “The number of people using our products in the UK has increased fourfold in the past year. It’s exciting to see them using AI to save time, increase productivity and get more done.”
AI adoption in the UK has exploded. I’m living proof of that. I’m increasingly consulting and building software for small and medium businesses, based on GPT models, for everything from managing workloads and automating proposals to making sense of data. This new deal simply formalises what’s already happening, with a shiny layer of sovereignty and a promise of safety.
But who chooses what’s safe? OpenAI? Big government?
If the furore surrounding digital ID has shown us anything, it’s that while the great British public have grown apathetic in some areas of their rights and freedoms, their data is not one of them. People care.
Hosting data locally is a win, but OpenAI still controls the models, the training and the policies. The deal may protect our data, but it doesn’t necessarily stop our over-dependence. Once public services start running on proprietary AI systems, switching away becomes very difficult, as anyone who’s ever tried to migrate away from Microsoft, Google or Salesforce will tell you.
So yes, data sovereignty matters. But so does the rest of the conversation. That stuff we put a pin in earlier? It’s not by the by, it’s the main conversation, and we need to address it sooner rather than later.
The advent of any new technology like AI will always come with ethical conundrums and questions that need answering. But do we keep the train at the station and wait for the track to be built, or do we chug along cautiously, build as we go and trust the path will become clear further down the line? What’s the risk versus reward here?
The Prize and Price of Productivity
The numbers are huge. The OECD estimates AI could add up to £140 billion to UK output each year by 2030. The government says it’s upskilling 7.5 million people, that’s one in five of the workforce, to use AI in their day-to-day roles. There’s talk of AI “Growth Zones”, AI sector champions, and the newly announced “AI Growth Lab”, which will test where red tape might be slowing innovation.
This is the biggest tech emergence since the internet. The race is on to make the UK an AI leader. But how involved the government is in that, and what it means for the future sovereignty of our data, remains to be seen.
Is “keeping AI data in the UK” code for handing it over to our government? In short, no, not at this time. But it’s something we’ll have to keep a close eye on.
The good news is, companies like OpenAI are having these types of conversations with governments across the globe. They’re well practised at pushing back on overreach, even if they haven’t got it down to a fine art just yet.
Why Sovereignty Matters
Data residency, the ability to store and process information within the UK, could finally unlock AI adoption for industries that have been hesitant.
With OpenAI servers based here, there’s less reason for those sectors to sit on the sidelines.
This could be the thing that gets the NHS using AI ethically and effectively, that lets teachers harness AI tools for planning without breaking GDPR, that helps legal firms work faster while keeping client data secure.
It’s the infrastructure piece we’ve been missing, and for that, credit where it’s due.
But this isn’t just about technology. It’s about trust.
Britain has spent the better part of two decades outsourcing everything: energy, housing, data, even justice. The fact that the government is now making a big deal about “sovereign servers” is kind of telling, about how far we’ve drifted from actual sovereignty. That’s not all bad, but if the UK is serious about using AI for public good, this deal could be the start of something important, provided it comes with transparency, independent oversight and a commitment to open frameworks that work for the betterment of everybody.
It can’t just be a handshake between ministers and big tech that fucks us out of our data, our rights or our privacy.
AI has the potential to level the playing field. Let’s make sure it remains a tool for me and you to do great things.