Briefing: AI needs to supercharge better regulation

The AI revolution continues apace, with optimists and pessimists once again locked in intellectual combat over the prospect of an Anthropic IPO.

Meanwhile, practical uses for this magical technology abound. Few stand out more clearly than compliance. Yet it’s the FCA we should be looking to for a lead.

The FCA is already proactive in adopting AI and exploring its use within regulatory workflows. The Handbook API enables regulated firms to access regulatory rules in machine-readable form and use AI to support compliance.

Jessica Rusu’s roadmap includes a first-responder agent for small cases, as well as future anti-money laundering (AML) workflows.

We need to rapidly advance this future and tie it to the vision outlined in the Mills Review earlier this year – one of proactive, real-time agentic interaction.

There’s a future where every regulated entity has a data graph and a cluster of compliance agents, configured and managed by humans. One agent will manage the relationship with the FCA’s master case management AI, which we can call Smith.

Your organisation and Smith will talk regularly, logging and resolving routine cases with barely a human in the loop. Reporting will move towards real time. And with the right data feeds and the PRA involved, monitoring insurers’ solvency ratios could too.

Think Open Banking for regulatory data.

Compliance reporting would move from thud to hum – and consumers would be better for it.

Mountains of data

The FCA, you see, is sitting on decades of supervisory and enforcement records, alongside evidence from the people it protects and the firms it regulates. Two streams leading to one ocean.

The bigger prize for our regulator and AI is not internal productivity – it’s at the intersection between the FCA and firms.

That history could make something like pre-emptive regulation possible. Smith could get in front of a case, approaching a firm’s agent when the data shows a familiar pattern emerging.

A cluster of complaints that looks like the early stages of a problem the FCA has seen before might, for example, prompt a conversation while there’s still time to put things right.

Complaints themselves could be handled in real time and more rapidly aligned with the Financial Ombudsman Service (FOS). Again, the data is all there.

It would be fantastic to see the first versions of this in the final quarter of 2027, provided the regulator tracks and stretches the technology as effectively as the firms it regulates. It seems to be on a faster track than regulatory pessimists might expect.

Regulation is about standards and we all rightly – and by obligation – respect and adhere to that mindset.

Compliance from the top down should enable progress. Regulation should also constantly appraise itself and strip out requirements that create unintended consequences for competition.

Now it can.

After all, every hour a regulated firm spends on compliance is ultimately a cost borne by the policyholder. 

We often hear on the conference circuit how far insurance is behind our siblings in banking. But, for many products, we have the same regulator.

Perhaps those who led insurance’s stalled efforts to follow Open Banking and build a market around common APIs deserve another swing at the plate.

A fresh push for Open Insurance feels about as likely as a Blueprint III announcement from Patrick Tiernan at Lloyd’s. So, perhaps openness emerges from the core.

The industry should ask for a new agent-to-agent relationship with the FCA. Call it Open Regulation.

Once firms have a common way to talk to their regulator, they have a reason to use the same standards with one another. Everything else can flow downstream from there.

If you were building a new regulated entity from scratch tomorrow, this is the future you would point your technology stack towards.

They’ve made tax digital. Now it’s time to make regulation digital too.