A new model for policing data: A decentralised data mesh approach
Over 90% of crimes are now estimated to involve a digital element, yet policing data is "often fragmented across 43 forces and recorded in different formats," according to the UK Home Office white paper From local to national: A new model for policing. The reforms it proposes, from fewer, larger forces to a National Police Service and Police.AI, all depend on joining that data up. AI, the paper says, offers new opportunities "only if the underlying data is accessible, consistent and well managed." How policing integrates its data nationally is an architectural decision, and it is being made now.
This paper argues that policing should connect its data rather than centralise it, and makes the case for a data mesh. Data would stay where it is generated, owned by the force or national body that produces it, and be published as reusable data products that investigation, analytics, and AI capabilities reach through distributed search and retrieval. The familiar alternative, a large central programme that copies force data into one national platform, risks rebuilding today's problems on a larger scale.
Highlights
- Compare centralising police data with connecting it, and see why large central data programmes have tended to fail.
- Explore how a data mesh could let forces collaborate without migrating data, and give national reach without a national copy.
- See why ownership, data protection, accountability, national standards, and security can stay where the data sits.
- Consider what Police.AI needs from the data beneath it, and why AI agents should have the same access controls and audit as people.
- Weigh three principles that would keep the reform's options open and its costs down.