The Home Office has updated its factsheet on police use of AI and the new National Centre for AI in Policing, PoliceAI

Key takeaways

• The factsheet is a policy summary. It states that Parliament sets the legal framework, and it names the Equality Act 2010 and the Data Protection Act 2018.
• Individual Chief Officers decide when and how their force uses AI, and they answer to their locally elected Police and Crime Commissioners.
• PoliceAI would test and benchmark AI tools and publish guidance. The factsheet says it will not make ethical judgements on whether a tool should be used.
• The Home Office says it intends to launch a pilot of the public-facing register of police AI by the end of 2026.

What is the Home Office factsheet, and is it binding?

It is a Home Office page on GOV.UK, first published on 9 June 2026 and last updated on 22 October 2026. The page summary says it summarises Home Office measures to support police forces in England and Wales to adopt AI responsibly. The factsheet itself does not state a sanction. The update log says only "The factsheet has been updated", so the text does not show what changed between June and October.

The factsheet states a policy position that where police forces use AI, it should be deployed responsibly, and that this means deployments must be lawful, ethical, transparent and based on "a robust assessment of algorithms before use". That is the government's stated expectation, not a statutory test. Government-funded guidance to help forces meet it is published on the College of Policing website, according to the factsheet, and it says PoliceAI is responsible for keeping that guidance up to date.

Who does the factsheet say is responsible for police AI?

In the factsheet's words, "Individual Chief Officers are responsible for deciding when and how to use AI within the police force they lead." They are accountable to locally elected Police and Crime Commissioners, and the factsheet says this will remain so until those positions are abolished and replaced.

On the law, the factsheet says Parliament sets the legal framework and that police must comply with all relevant laws and regulations, including the Equality Act 2010 and the Data Protection Act 2018. It says Home Office grant funding for AI adoption is primarily provided under Section 57 of the Police Act 1996. It also says the government does not currently support police using AI to answer 999 emergency calls.

For facial recognition, the factsheet points to a public consultation the government launched in December 2025 on a new legal framework for law enforcement use of biometrics, facial recognition and similar technologies. We cover the regulator's findings separately in our article on the ICO outcomes report on five police forces.

What will PoliceAI do, and what will it not do?

The factsheet describes PoliceAI as a national capability hosted by the College of Policing on behalf of policing in England and Wales. It says the centre began mobilising in April 2026, was formally launched in June, and will transition into the National Police Service once that service has been set up. It will be funded through a grant agreement issued by the Home Office under Section 57 of the Police Act 1996.

Three of the functions it lists are these. A new AI Lab, which the factsheet says assesses commercial products and open-source models, will make assessments through testing, validation and benchmarking and will run a sandbox where policing can test models on data before operational use. An AI enablement function will support forces to adopt tools the Lab has developed, tuned, tested or assured, including funding for set-up costs. A strategy, oversight and coordination function will provide public-facing communications, including publishing and maintaining a registry of models used by police in operational settings.

The limits are stated just as plainly. PoliceAI will not host AI models for operational policing, will not run cyber and information security assessments of models, and will not be asked to focus on commercialising its products or services. The factsheet adds: "It will not make ethical judgements on whether an AI tool should be used." Operational decisions on AI output stay with Chief Officers.

The factsheet also says PoliceAI hosts the "Policing AI Threat Hub" (PATH), which works on criminal misuse of AI, including protecting the criminal justice system from AI-generated false evidence. In its first year, the stated focus includes case file assistants, disclosure assistants, crime data integrity tools, rapid analysis of CCTV and digital media, and image identification and classification.

What funding and transparency commitments does it describe?

The factsheet says the Home Office has provided more than £50m in direct grant funding since the 2024 election for specific projects, such as new Live Facial Recognition vans. It says the Police Reform White Paper announced a further £115m for police adoption of AI and automation over the next three years, and that a separate £26m for a national facial recognition system takes the White Paper package to £141m. These are the government's own figures as stated in the factsheet; we have not checked them against the White Paper.

On transparency, the factsheet says the White Paper announced a public-facing registry of AI deployed by police forces. It says the registry will set out what each tool does, why it is used, the risks identified and how they are mitigated, along with technical details, and that a disclosure policy is still being developed with policing leaders. It adds: "The government expects all Chief Constables to complete the registry once it is launched and will act where this is not the case." A pilot of the public-facing element is intended by the end of 2026, per the factsheet. That is an intention, not a delivered system.

For comparison with the data protection track, see our article on the Data (Use and Access) Act automated decision-making regime, which applies to a different set of decisions.

What should lawyers and privacy counsel do now?

The factsheet gives no instruction to lawyers. Its practical use is as an official statement of the accountability chain and of what a national centre and register are meant to produce.

• Criminal practitioners: when police AI tools appear in a case file, ask which force deployed them and who made the decision to use them, since the factsheet places that decision with the Chief Officer.
• Public-law and privacy counsel: separate the register (which the factsheet describes as not yet launched) and the AI Lab's stated functions, and the proposed facial recognition legal framework, which the government consulted on from December 2025, from anything already in force, and cite the factsheet as a policy statement only.
• Advisers to AI vendors: note that PoliceAI is described as a testing and benchmarking route and that the factsheet says PoliceAI will not make ethical judgements on whether an AI tool should be used.
• All readers: watch the GOV.UK page for further updates, since the update log shows two entries so far.

What are the key dates and parties?

What we did not verify

What we opened: the GOV.UK publication page and the full text of the accessible HTML version of the factsheet, on 2026-10-25. The page dates come from the GOV.UK update log.

What we did not open: the Police Reform White Paper, the College of Policing guidance and governance pages, the December 2025 facial recognition consultation, the PoliceAI contract with Deloitte, and any earlier version of the factsheet. We therefore cannot say what the 22 October 2026 update changed, and we have not checked the funding figures against the White Paper.

What we refuse to claim: that the factsheet is binding, that it creates rights or duties, that the public register is live or how far the AI Lab's functions are operating, or what any force currently does with AI beyond what the factsheet says.

Related briefings

• AI Regulation News hub
• Europe AI regulation tracker
• Attorneys and AI regulation
• ICO outcomes report on facial recognition in law enforcement across five forces
• Sweden: police real-time AI facial recognition law
• UK Data (Use and Access) Act automated decision-making regime
• ICO statutory code of practice on AI and automated decision-making