Tom Loosemore

AI agents will join up government before government joins up government. This will be a mixed blessing for citizens, with big implications for trust, equity and accountability.

Let’s say you move house in the UK, and want to inform relevant public services.

Well, good luck. You’re on your own.

There’s no single, joined-up place to tell all public services that you’ve moved, not even just for central government services – hell, not even just for DWP’s services. Shameful. Even the French have fixed this.

Except we’re not on our own any more.

The AI agents are coming. They’ll do the joining up of public services for us, for good or for ill. It might take a few years, but they’ll get there sooner than government will itself.

This morning I’ve been playing with using an AI Agent (Claude Cowork) to do the hard work of updating my address across all public services which matter to me. I linked Cowork to my email archive, then gave it this prompt:

“I’d like you to check through all my email going back several years to find valid messages from any of these services. Then create me a tool to update such services with my new address.”

Have a look at the screen grab below, and remember this is the worst it will ever be. With one further prompt it added my bank, utilities, Amazon etc.

What it won’t do – at least at present – is log into these services and update your address on your behalf. That’s not because it couldn’t. Technically, it’d be trivial.

Rather, it’s because the makers of Claude Cowork have decided that to do so is currently too much of a security risk for customers. OpenClaw would have no such scruples.

Why does this matter? Well, let’s fast forward to a future where all your government interactions are intermediated by your AI Agent.

Firstly, they won’t be ‘your’ AI Agent. They’ll likely be US-owned, with associated sovereignty risks.

Many AI Agents won’t be free, cue concerns around equity and access. And ‘free’ ad-supported AI Agents will bring their own problems around trust. I spent years failing to stop Google selling search results to scammers mimicking GOV.UK services.

And how will AI Agents be held accountable for their actions, should you suffer as a consequence of it getting things wrong?

Finally, what of the legitimacy of the state if you do not even recognise that you’re interacting with tax-supported public services?

So what would I do, were I in government?

I’d explore restricting access to only those AI Agents who sign up to a GOV.UK kitemark, with legally-mandated conditions to manage the risks above.

India used similar regulatory robustness before Google Wallet could host Aadhaar credentials.

I’d experiment with that kitemark right now, to try to shape things while they’re still somewhat malleable. That window won’t remain open for long.

There are excellent people inside GDS who understand all this. Let’s hope they get their voices heard.

This is what global digital leadership looks like.