Daily Update, 21 October 2026: Home Office says a version of a public police AI tools registry is expected by the end of 2026

The Home Office says "A version of the public registry is expected to be published by the end of 2026", referring to a registry of AI tools used across policing, according to written answers from Sarah Jones, answering for the Home Office, published on Tuesday 20 October. Also today: a planning ministry answer that says its AI planning-decisions trial used no control group, what the government says about £3.6 million for AI projects at regulators, and a pointer to the new Digital and Technologies Jobs Plan.

Police AI: a registry "by the end of 2026", and an advisory group to "consider" benchmarks

Independent MP Iqbal Mohamed tabled two questions on police AI oversight. Both received the same answer from Sarah Jones for the Home Office on 20 October.

• The registry: "PoliceAI will also publish and maintain a registry of AI tools in use across policing. This is being developed in partnership with academics and is currently in testing with forces. A version of the public registry is expected to be published by the end of 2026."
• What it will hold: "The registry will include details on the steps taken by the force to deploy the AI responsibly, including ethical assessment and accountability arrangements."
• Benchmarks: "This year, PoliceAI will establish an independent advisory group, that the APCC will be a member of, to advise on AI ethics at a national level and consider nationally consistent accuracy and bias benchmarks for AI tools." The APCC is the Association of Police and Crime Commissioners. The wording is that the group will "consider" benchmarks, not that it will set them.
• Who decides locally: "Procurement and deployment of AI at a local force level is the responsibility of operationally independent Chief Constables, accountable to their locally elected Police and Crime Commissioners (PCCs)."

The first question asked how many forces in England and Wales have a police ethics board or equivalent body, and how many have a PCC who has formally scrutinised the force's use of AI in the last 12 months. It also asked whether the Home Secretary plans to mandate a minimum standard of ethics board coverage before the abolition of PCCs in 2028, as the question put it. The answer, which we read in full, gives no count and does not say whether such a standard will be mandated. On the transition, it says officials are working with partners "to support and oversee transition to new police governance arrangements in May 2028".

On timing: when the Home Office announced the launch of PoliceAI on 10 June, its news release said the registry would be "developed in partnership with CENTRIC at Sheffield Hallam University" and that "A first version will be available by the autumn." In her launch speech the same day, Sarah Jones said PoliceAI would produce it "by, I hope, the autumn." On our reading, the target in yesterday's answer, the end of 2026, is later than the autumn date given in June.

AI planning tool: "A control group was not used"

Asked by Conservative MP Dr Ben Spencer what the trials of the AI planning tool found, whether a control group was used and when a national rollout will begin, Matthew Pennycook, answering for the Ministry of Housing, Communities and Local Government, said the first testing phase of the "AI augmented planning decisions prototype, which was undertaken in three local planning authorities (LPAs), produced promising results."

• Method: "A control group was not used."
• Next phase: "Testing has now progressed to 12 further LPAs."
• Rollout: "No final decision has been made in respect of a national rollout." The answer describes the approach as "exploratory and evidence-led, to ensure any future adoption is safe and ethical".

The answer gives no figures for the "promising results" from the first phase.

Regulators: £3.6 million to eight AI projects

An answer from Chris McDonald, for the Department for Business, Innovation, Science and Trade, also to Dr Spencer, says: "The Regulatory Innovation Office awarded £3.6 million through the AI Capability Fund to eight regulator-led projects to strengthen capability to regulate and use AI".

• Areas covered: "digital regulation, energy, nuclear safety, aviation, data protection, autonomous vehicles and medicines".
• Status: "Seven completed funded delivery by 31 March 2026, with one still in active delivery."
• Pioneer Fund: separately, round 4 of the Regulators' Pioneer Fund "awarded up to £7.9 million to 14 projects testing innovative regulatory approaches across health, housing, transport and environmental regulation", and "Delivery is ongoing".

The question asked what each project delivered. The answer lists the sectors covered but does not name individual projects or say what each one delivered.

Also: Digital and Technologies Jobs Plan

DCMS published its Digital and Technologies Jobs Plan on 21 October. It says "Government and employers commit to around 20,000 entry-level jobs a year through a single Skills Deal for employers", a figure the plan also calls "an ambition shared by government and industry, not a forecast", and aims to reach "one million young people with meaningful AI and technology learning through TechFirst by 2028", prioritising the most disadvantaged schools. We have published a fuller guide to the plan.

Why it matters

On our reading, yesterday's answers describe public AI tools that are still being tested, and the evaluation details are thin in places. The planning prototype moved from three planning authorities to 12 more without a control group in its first phase. The police registry is "currently in testing with forces", with a version "expected" by the end of 2026, and the national advisory group will "consider", rather than set, accuracy and bias benchmarks. The police answer did not give the counts its question asked for, the planning answer did not give figures for its results, and the regulator answer did not give the project-by-project detail requested.