REQUEST
I am writing under the Freedom of Information Act 2000, to make a request for information relating to your organisation’s development and/or use of digital decision tools for policing and law enforcement.
By digital decision tools, I mean any technology that uses software algorithms to process digital data and generate outputs either to (a) *support* policing decisions, by providing information or recommendations to a human police officer, who retains discretion in making the final decision; or (b) *automatically* trigger a policing or law enforcement decision based on the generated output(s).
By ‘policing and law enforcement decisions’, I mean any decisions taken by police officers in carrying out their duties to detect and prevent crime, to protect the public and reduce harm, to maintain public order, to enforce the law, and to bring offenders to justice. This may include (but is not limited to) decisions concerning the imposition of enforcement notices (e.g., speeding fines), case/complaints-handling, resource allocation, criminal investigations (particularly those concerning whether to pursue inquiries in relation to specific persons, and what action(s) to take, such as arrest or questioning, stop and search etc.), and decisions made at the pre-trial stage about individuals, including detention and charging decisions.
EXAMPLE TOOLS OF INTEREST
To help with my inquiry, examples of the kinds of tools I am requesting information on include:
- *Digital dashboards*: any tool which processes and organises data sets to generate data visualisations to assist policing work. For example, they might provide information about ongoing and/or recently reported incidents, the analysis of crime patterns, workload and resource allocation, and data shared with public agencies such as the emergency services. These dashboards might, for instance, be used to improve emergency response times, assist officer decision-making, resource management, and so on. For example, Qliksense.
- *Individual risk assessment tools*: any tool which processes data to generate predictions about the ‘risk’ posed by an individual relating to an undesirable outcome, such as their risk of committing a serious violent crime in the future, risk of general recidivism, risk of self-harm, or risk of becoming a victim of harm. For example, the London Gangs Matrix or the Harm Assessment Risk Tool (‘HART’) that has previously been used by Durham Constabulary.
- *Geospatial crime ‘hotspot’ mapping tools*: any tool which processes historical crime data to identify patterns indicating potential crime ‘hotspot’ areas (i.e., geographical locations where crime is predicted to be most likely to occur within a given time period). For example, PredPol.
- *Automated biometric analysis tools* (including live facial recognition systems): any tool which processes biometric data for the purposes of identifying individuals ‘of interest’ (i.e., according to s.205 of the Data Protection Act 2018: biometric data is data relating to the physical, physiological, or behavioural characteristics of a person, such as facial images, fingerprints, or retina scanning). For example, NeoFace Watch.
- *Any other tool to assist policing and law enforcement work*: any other tool that uses software algorithms to generate outputs to help inform policing decisions made by your organisation, or to automatically trigger decisions based on those outputs (e.g., Automated Number Plate Recognition (ANPR) systems, social network analysis, case ‘solvability’ analysis, and so on).
--------------------------------------------------------------------------------------------
SPECIFIC INFORMATION REQUESTED
My specific information requests are listed and numbered below.
To help reduce the resources needed to respond to my numbered requests, you may limit your detailed responses to *ten* tools, and then simply list the name(s) and policy purpose(s) of any other tools.
If your organisation has deployed and/or developed *ten or fewer* tools in total (including those currently deployed, in development, being piloted or formerly developed or piloted but never deployed), please skip straight to the numbered requests below.
If your organisation has deployed and/or is developing (or has developed) *more* than ten tools in total, please follow the guidance below to identify which ten tools to provide information about, in response to the numbered requests:
- If *more* than ten tools have been deployed for *operational* use by your organisation, please respond to the numbered requests for the ten *most recently* deployed (including those no longer in use) and merely list the name(s) and policy purpose(s) for any other tools that have been deployed or developed (including ones currently in development);
- If your organisation has deployed *fewer* than ten tools for operational use (or none at all), please respond to the numbered requests in relation to any other tools that your organisation is (or was formerly) *developing or piloting, * until you have reached a total of ten tools. Again, please respond for those tools *most recently* put into development or piloted until you reach the limit of ten tools in total, and then simply list the name(s) and policy purpose(s) for any remaining tools.
FOR EACH RELEVANT TOOL (UP TO TEN IN TOTAL), CAN YOU PLEASE:
- Supply the *name(s)* of each tool, including any tool(s) which are no longer used (or were developed and/or piloted but never deployed).
- Please specify the stage of maturity of each tool in terms of the following three possibilities (i) currently in operational use or formerly deployed for operational use (ii) currently under development or was developed in the past but never deployed (iii) currently being piloted or was piloted in the past, but never deployed.
- Explain the *reason(s)* why your organisation decided to develop and/or deploy each tool, including the intended *policy purpose(s)* of the tool (e.g., to reduce criminal offending by aiming to improve the identification of potential victims of a criminal offence or to identify individuals ‘of interest’ to police, using the automated analysis of digital data).
- Identify the data sources used as ‘inputs’ to the tool and indicate whether this data has been collected from internal police data sources (e.g., crime data), external public-sector data sources (e.g., data held by local authorities), or external data sources made available by private-sector organisations (e.g., mobile communications data). Please further indicate whether the tool is updated with new data on a manual basis, by a human user responsible for inputting data (including how frequently this usually occurs), or if it is updated on a real-time basis using an automated live data feed.
- Identify the *output(s)* that each tool produces/produced in order to serve its intended policy purpose, and what these outputs are taken to *indicate or signify.* For example: an individual risk assessment tool may produce a ‘risk classification’ for each assessed person (e.g., ‘high’, ‘medium’ or ‘low risk’). What is it, specifically, that each person is at ‘high risk’ or ‘low risk’ of, in this scenario (e.g., being arrested under suspicion of having committed a serious offence within the next two years)?
- Indicate the kind(s) of *decision(s)* each tool supports/supported or triggers/triggered and about *whom or what*? For example: who to stop and search? Where to send police patrols? Should an arrested person be retained in police custody?
- Identify the *intended user(s)* of each tool. For example: does/did the tool provide information to a front-line officer on patrol, a duty officer in a custody suite, both, or someone else?
- *List* and provide electronic copies of any *additional documents* or information relating to each tool, including:
- Any training manuals or materials for staff, promotional materials, briefing papers, reports, or records of performance evaluations and other technical documents that provide overviews or summary information about each tool.
- Any impact assessments (e.g., data protection or human rights impact assessments), or risk assessments, that have been undertaken in relation to the tool or tools indicated.
- Any legal documents relating to the assessment or statement of the legal authorisation/legal basis for using each tool.
FOR TOOLS *CURRENTLY* OR *FORMERLY* IN OPERATIONAL USE ONLY, CAN YOU PLEASE:
- Please indicate whether in relation to each tool, users receive any *training* about tool-use, indicating what that training consists (or consisted) of, who provides this training (including whether this is in-house or external), and who receives training.
- Indicate the *time-period* during which each tool was, or had been, in operational use, identifying the *date(s)* when each tool was first deployed (and when its use was ended, if applicable).
- If your organisation later decided to *stop* the tool’s operational use, can you please further explain your reason(s) for doing so, and indicate when that decision was made?
FOR TOOLS CURRENTLY UNDER DEVELOPMENT OR PILOTING (OR FORMERLY UNDER DEVELOPMENT OR PILOTED, BUT NEVER DEPLOYED) ONLY, CAN YOU PLEASE:
- Indicate the *time period* during which development or piloting for each tool took place/has been taking place, including when development or the pilot began, the *expected/planned launch date* for tools still under development, and whether the launch will be on a pilot basis or for operational use.
- If your organisation later decided to *abandon the development* of a tool or decided *not to deploy* that tool for operational use following a pilot period, can you please further explain your reason(s) for doing so, and indicate when that decision was made.
RESPONSE
FOLLOW UP REQUEST
To summarise, my refinements include:
- Removal of my request for “electronic copies of any *additional documents* or information relating to each tool.” This request may be particularly onerous in terms of time needed to comply.
- Limiting the scope of my request to those tools that are *currently in operation only*, again with a view to reducing the time required to identify, locate, and extract the relevant information. You can limit the number of tools to ten, chosen on the basis of those that have *most recently been deployed.* This will reduce the number of tools within scope, and remove the need for you to respond to requests 11 and 12.
- Limitation of the scope of my request to the *last five years.*
- A significant reduction in the scope of example tools, to now focus on digital dashboards,individual risk assessment tools, geospatial hotspot mapping tools, automated biometric analysis tools ONLY (i.e., no general request for other tools).
In case it helps, other forces have found it useful to reply using a simple table which sets out their responses on a tool-by-tool basis.
I hope that my suggestions can help us to find a productive compromise, and I appreciate the time you have taken to consider and respond to my request.
RESPONSE