Guidance

Police use of artificial intelligence (AI): factsheet (accessible)

Published 9 June 2026

What is the government’s overall policy position on police use of AI?

• In the government’s AI Opportunities Action Plan, the Prime Minister set out his intention to maximise the potential for AI to make the public sector more efficient and effective and deliver the government’s Plan for Change.

• To date, most AI activity pursued by police forces in England and Wales has been undertaken by Chief Officers using their core funding and accountable to their elected Police and Crime Commissioners.

• Since the 2024 election this has been supplemented by over £50m of direct grant funding from the Home Office for specific projects, such as to fund new Live Facial Recognition vans.

• In the Police Reform White Paper, the government announced a further £115m for police adoption of AI and automation which covers a range of projects such as creating a new National Centre for AI in Policing (“PoliceAI”). The aim of these projects is to rapidly equip policing with high quality AI that can make the biggest difference to public safety outcomes in local communities.

• Where police forces use AI, it should be deployed responsibly. That means deployments must be lawful, ethical, transparent and based on a robust assessment of algorithms before use. It should be clear who is accountable for the AI’s performance and the use of its output by police personnel, and there should be clear operating procedures in place that set out what appropriate usage looks like. These principles are critical for building and maintaining public consent for the use of this technology.

• To assist the police in using AI responsibly, the government has funded the National Police Chief’s Council to develop guidance, which is published on the College of Policing’s website. PoliceAI will provide a range of services that assist police forces to comply with this guidance.

How does police use of AI support police reform?

• The Police Reform White Paper sets out the government’s intention to deliver a more capable police service, equipped with the latest technology to improve the prevention, detection and investigation of crime, to improve public safety outcomes at a local level.
• Investment in AI offers policing significant opportunities to improve the efficiency and effectiveness of their activity across many areas, improving police performance. It can also free up resources to take on other tasks, including policing neighbourhoods and engaging with victims and witnesses.

How is AI in policing defined?

• There is no universal definition of AI. Within policing, the definition adopted by the Police Chief Scientific Advisor is ‘a machine that learns, generalises, or infers meaning from input, thereby reproducing or surpassing human intelligence.’

• The UK Government definition is ‘an AI system is a machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments. Different AI systems vary in their levels of autonomy and adaptiveness after deployment.’

• AI relies on input data, an algorithm based on that data, and compute power to process that algorithm.

In what ways might the police use AI?

• In theory, policing could use AI to improve the efficiency and effectiveness of almost all their existing activities.

• In practice, when deciding whether to use AI, Chief Officers will need to make judgements that consider a wide range of factors including cost, model accuracy, user training, quality of source data, lawfulness, ethics and public acceptance.

• The following areas are considered high potential police use cases for AI by the government:

  • Handling information required for policing duties, for example as part of an investigation – triaging and analysing digital evidence, redaction of text and audio-visual files, summarising case files, assisting with disclosure obligations and detecting deepfakes. This can make investigations quicker as well as more thorough, allowing officers to locate and deal with key information in an investigation in minutes rather than days. This can provide a faster service to victims and lead to more early guilty pleas.

  • Identifying criminal behaviour, wanted suspects or missing persons – for example through facial recognition or automated numberplate recognition. There are many examples where use of AI for this purpose has led to the arrest of wanted individuals, including rapists and child sex offenders, making communities safer.

  • Improving operation of force control rooms – for example by supporting the triage of 101 non-emergency calls, so that those not for the police are dealt with by the appropriate agency, or by transcribing calls, so the police have a record of calls and can use it to improve their training and response. The government does not currently support police using AI to answer 999 emergency calls.

  • Supporting with time-consuming back-office tasks – such as classification of crimes, note taking, digitising paper files for cold case reviews, summarising documents, translating documents, transcribing witness statements and form-filling. This frees up police officers and staff for frontline responsibilities including patrols, investigations and engaging victims and witnesses.

Who is responsible for police use of AI?

• Individual Chief Officers are responsible for deciding when and how to use AI within the police force they lead.

• They are accountable to their locally elected Police and Crime Commissioners and this will remain the case until the positions are abolished and replaced.

• Parliament sets the legal framework governing the use of AI. When using AI, the police must comply with all relevant laws and regulations including the Equality Act 2010 and the Data Protection Act 2018.

• Where the Home Office provides grant funding to the police to support with the adoption of AI technology, this is primarily done through Section 57 of the Police Act 1996, which gives the Home Secretary the power to give financial assistance in connection with the provision or maintenance of “such organisations, facilities and services as she considers necessary or expedient for promoting the efficiency and effectiveness of the police.”

What measures have the government taken so far to support the police use AI?

• In 2024-25 and 2025-26, the government supported the police adopt AI with over £50m of investment. This included funding for:

  • Testing facial recognition algorithms and rolling out live facial recognition vans.
  • Subsidising police procurement and use of AI software enabling the redaction of audio-visual files.
  • Work to develop and test deepfake detection technologies
  • Research to understand public attitudes toward the use of AI by the police. This has since been supplemented by the work of organisations such as CENTRIC, who surveyed over 10,000 members of the public on AI in policing and published the results in October 2025.
  • Innovative trials through the Police Chief Scientific Advisor’s STAR fund, for example to test the use of AI to summarise case files or match footwear upper evidence to a national database.
  • A team to support the NPCC AI lead, Alex Murray, who published an AI Strategy and produced guidance documents for the police such as an AI Adoption Playbook. Alex Murray is now the interim Director of PoliceAI.
  • In December 2025, the government launched a consultation on a new legal framework for law enforcement use of biometrics, facial recognition and similar technologies.

What measures were announced in the Police Reform White Paper on AI in policing?

• The Police Reform White Paper sets out a plan for the government to invest £115m in policing over the next 3 years to deepen support for the responsible adoption of AI and automation technologies by the police.

• This includes funding:

  • For a new National Centre for AI in Policing (“PoliceAI”).
  • To continue work started in 2025-26 to support the police to more rapidly adopt AI-enabled audio-visual redaction tools and efficient processes, estimated to save the equivalent of 550 FTE per year if adopted by all forces in England and Wales.
  • To continue work started in 2023-24 to support the police use of robotic automation processing software. Funding to date (£10.1m) has supported 27 forces use this software. The White Paper funding will enable all forces to have self-sustaining robotic process automation capabilities by the end of the Parliament. This capability has many benefits, including improving the quality of police data as a precursor to AI deployment.
  • To continue work started in 2024-25 to assist the police use AI in triaging, transcribing and managing 101 non-emergency calls and to assist with police/public engagement more broadly.
  • For 40 new Live Facial Recognition units to assist the police in high crime hot spots and in policing town centres. This funding will also support a stronger national team to oversee national coordination of this capability.
  • The government is also investing £26m in developing and delivering a national facial recognition system, taking the overall White Paper investment package up to £141m for AI and automation.

What is PoliceAI?

• A new national centre for AI in Policing, which began mobilising in April 2026. It will be formally launched in June.
• PoliceAI will consolidate a range of activities on AI of relevance to policing into a new national team, meaning activities are undertaken once on behalf of all forces, bringing greater consistency across policing, and providing the public with greater reassurance that the police are using AI responsibly. This is in keeping with the White Paper’s ambition to create a more coherent, consolidated national approach to policing.
• PoliceAI will be administered by a team hosted in the College of Policing. It will transition into the National Police Service as soon as possible, once the National Police Service has been set up.
• PoliceAI will be funded through a grant agreement, issued by the Home Office under Section 57 of the Police Act 1996.

Why is the government setting up PoliceAI?

There are several reasons. The key arguments are:

• AI gives rise to new requirements that are best undertaken by a team once nationally on behalf of policing, rather than by each force independently (or not at all by forces due to local constraints). Example requirements include evaluating the accuracy of AI models that appear in different software packages, tuning AI models to work more accurately in a UK policing context or providing prompt guides for AI models used by multiple forces. National consolidation of activity on AI is consistent with the direction set within Police Reform White Paper and the approach elsewhere in the public sector, such as with the Defence AI Centre.
• National delivery increases the prospect of a consistent AI-enabled service being provided to the public across all police forces. Consistency is vital to deliver higher standards, improved police performance and better outcomes for local communities.
• Consolidating responsibilities on AI reduces the complexity of the current national capability delivery landscape, which can make it difficult for individual forces, academics and the private sector to know who to engage with. Further consolidation will be achieved through creation of the National Police Service and PoliceAI is a first step towards that, which the government is delivering now because it does not require new legislation.
• Backed by a significant increase in funding, PoliceAI will be able to undertake evaluations and benchmarking exercises that can give both Chief Officers and the public confidence that the AI in use by the policing is accurate and bias has been minimised.
• The White Paper also set out plans to equip policing with the latest technology. PoliceAI will support police acquisition and use of AI.

What will PoliceAI do?

• The government will charge PoliceAI with focusing on supporting the police rapidly but responsibly adopt AI in the areas of policing business most likely to be effective in: preventing and detecting crime, catching wanted suspects, finding missing people, speeding up and enhancing the quality of investigations, improving the experience of victims and witnesses and making the police more productive, freeing up officers from support services to be on the frontline.
• By focusing on these areas PoliceAI will make improvements to public safety outcomes.

• To achieve this, the national centre will be funded to administer a range of functions on behalf of policing. These include:

  • A new AI Lab focused on developing, testing and assuring AI tools, products and services. It will assess commercial products and open-source models for their applicability and performance in a UK policing context, taking steps to increase accuracy and usability (e.g. by building user interfaces, tuning models or providing prompt guides) of these models. The Lab will make informed assessments about AI models and tools through testing, validation and benchmarking, allowing Chief Officers to make more informed procurement and deployment decisions. The AI Lab will run a ‘sandbox environment’ that allows policing to test and adjust models on data before they are deployed into an operational environment. The AI Lab will also seek to maintain an up-to-date evidence base on AI advances, so policing can continually exploit the technology, liaising with other organisations with expertise in this area like the AI Security Institute.

  • An AI enablement function focused on supporting forces take practical steps to adopt AI models and tools, which the AI Lab has developed, tuned, tested or assured. This offer will include the provision of funding to forces to get over the initial financial burden of set-up and training costs, working with BlueLight Commercial on national procurements to achieve economies of scale, connecting policing to academic leaders on AI, the collation, maintenance and publication of guidance and supporting forces with business change. The national centre should be seen as the first port of call for i) police forces when seeking practical resources, advice and guidance on exploiting AI models, ii) AI vendors, academia and other law enforcement entities seeking to engage policing on AI and iii) government where it seeks to bring together a policing view on AI issues, such as on new legislation.

  • A strategy, oversight and coordination function. This will provide public facing communications on policeAI use, including publishing and maintaining a registry of what models are being used by the police in operational settings and the checks and balances undertaken before deploying them, assessing the priority use cases to improve with AI, acting as a centre of expertise on criminal misuse of AI, and engaging with comparators across the public sector like Justice AI and the Defence AI Centre; as much as possible PoliceAI should re-use AI products developed elsewhere rather than starting from scratch.

What won’t PoliceAI do?

• It will not be asked to focus on commercialising its products or services. Its sole focus must be on improving public safety outcomes via the responsible use of AI by the police.

• Creating new AI models from scratch for policing will not be a priority and only pursued by exception where use cases/circumstances are unique to policing, cannot be addressed by open source/procurable models and where there is a strong value for money case to be made.

• PoliceAI will not host AI models for use in operational, day-to-day policing activity. It is expected that most AI models will be hosted and run by individual forces, or other available national services.

• It will not undertake cyber and information security assessments of AI models. The Police Digital Service already receives funding from both the Home Office and force subscriptions to do this.

• It will not make ethical judgements on whether an AI tool should be used. Nor will it take operational decisions on the output of AI models. This responsibility will remain with Chief Officers, with PoliceAI providing the resources and evidence to assist with making informed procurement and deployment decisions.

What uses of AI will PoliceAI focus on?

• In its first year, it will focus on areas of policing where the AI is sufficiently advanced to make a major impact on public safety outcomes for local communities. This includes:
• Accelerate delivery of the following use cases: case file assistants, disclosure assistants, crime data integrity tools, the rapid analysis of CCTV and digital media, and image identification and classification. We expect all of these to not only reduce administrative burdens on forces, but also to free up capacity to follow new evidential leads or be redeployed to other frontline duties.
• Full exploration of Copilot capability and establishment of appropriate governance and safeguards.
• Piloting of transcription and translation tools for wider rollout in future years across all policing use cases.
• Exploring the potential for AI to address retail crime and tool theft, for example by matching reports of stolen items to online sale sites.
• Subsequent priorities will be decided on an annual basis, following dialogue with Chief Constables.

Will PoliceAI also look at the growing threat of criminal misuse of AI?

• PoliceAI will contain a “Policing AI Threat Hub” which will help build a consistent national response across local policing to criminal misuse of AI. This will include improving training and guidance, better recording of AI-enabled crime, and action to protect the Criminal Justice System from the threat of AI-generated false evidence.
• It will also look to equip police forces with reliable, benchmarked tools to support the detection of AI generated deepfakes.
• The government takes the issue of criminal misuse of AI very seriously and has already created several new criminal offences to address this novel and growing threat, particularly its impact on violence against women and girls. From May 2026 it is illegal to create, adapt or supply a tool which generates non-consensual intimate images, and it is already a crime to create, share or requesting non-consensual intimate images, including when those images are AI-generated or deepfake. These offences will cover not only the users of these apps but also the developers who will now be criminally liable and could face a prison sentence and/or a substantial fine. We have also introduced similar legislation to ban generators of AI child sexual abuse material, the first of its kind in the world. The government has gone further and introduced an amendment to the Crime and Policing Bill to allow government to close gaps in the Online Safety Act, and bring unregulated AI services, including chatbots like Grok, into scope. This will empower regulators to ensure that criminal AI tools and illegal AI content are no longer appearing on online services.

How will PoliceAI be governed?

• It will operate to the terms of a grant agreement issued by the Home Office to the College of Policing under Section 57 of the Police Act 1996. This will set PoliceAI’s annual objectives, ensuring they align with the Home Secretary’s priorities for the police.
• The national centre will be run by a Director, recruited through fair and open competition. The Director will ultimately be accountable to the CEO of the College of Policing and the College’s Board for their performance who in turn will be accountable to the Home Secretary.

How will PoliceAI interact with other organisations in the policing ecosystem?

• The end goal is for PoliceAI to be part of the National Police Service. Prior to the set-up of the National Police Service, it will need to collaborate with other organisations within the policing ecosystem. In particular, PoliceAI will need to work:

  • With other teams in the College of Policing, who lead on the development of standards for policing to follow through Authorised Professional Practice documents, such as on data-driven technologies, and who have a range of existing resources that can be leveraged such as the What Works Centre and Centre for Police Productivity.
  • The Police Digital Service, who lead on cyber and information security for national policing (including for AI models) and who can host/run AI models nationally.
  • The National Data & Analytics Office (the NDAO), who provide a centralised accessible national data and analytics service for policing. This service covers responsible use of data and tools that enable insight-led decision-making. The NDAO will deliver a National Data Integration and Exploitation Service (NDIES) for the police, which will provide practical guidance to forces seeking to procure data integration tools, develop national data integration standards and deliver a national solution to integrate and exploit data at a national level within the cloud. The AI centre will work closely with NDAO where AI models are used for national level data exploitation.
  • The Office of the Police Chief Scientific Advisor which provides independent, expert scientific advice to the policing sector.
  • BlueLight Commercial who lead on commercial activity for policing at a national level.

How can I find out more about PoliceAI?

• PoliceAI will formally be launched in June.
• We are now working closely with the NPCC AI portfolio and the College of Policing to set it up. This includes recruiting its staff, agreeing its priorities, finalising its funding and determining an approach to monitoring and evaluation.
• A public facing website will also be created.
• The Home Office is not currently partnering directly with any private sector company, not for profit or any other non-UK public sector entity to deliver the national centre. The NPCC AI portfolio has been working with Deloitte to develop an operating model for the organisation.

What is your policy on facial recognition?

• The government launched a public consultation in December 2025 to develop a new legal framework for law enforcement use of biometrics, facial recognition and similar technologies.
• We want to ensure all police forces across the country can use facial recognition technology and other similar technologies with greater confidence and that its uses and limits are clear to the public.
• We have published a factsheet on police use of facial recognition including examples of positive results from using the technology, including the arrest of violent criminals, burglars and rapists.

When will you publish a register of policeAI use?

• In the Police Reform White Paper, the government announced that through the National Centre for AI in Policing, we will create a public facing registry of the AI being deployed by police forces, including details on the steps they have taken to ensure the reliability of tools before being used for operations.
• The registry will set out what each AI tool in use by the police does, why it is used, the risks identified and how these risks are mitigated. A policy on what should be disclosed is currently being developed and agreed with policing leaders, it is expected to align with FOI exemptions.
• The government expects all Chief Constables to complete the registry once it becomes available and will act where this is not the case.