Quickly exit this site by pressing the Escape key Leave this site
We use some essential cookies to make our website work. We’d like to set additional cookies so we can remember your preferences and understand how you use our site.
You can manage your preferences and cookie settings at any time by clicking on “Customise Cookies” below. For more information on how we use cookies, please see our Cookies notice.
Your cookie preferences have been saved. You can update your cookie settings at any time on the cookies page.
Your cookie preferences have been saved. You can update your cookie settings at any time on the cookies page.
Sorry, there was a technical problem. Please try again.
This site is a beta, which means it's a work in progress and we'll be adding more to it over the next few weeks. Your feedback helps us make things better, so please let us know what you think.
Freedom of information request reference no: 01.FOI.23.032794
I note you seek access to the following information:
I would like to submit a request for information regarding the Metropolitan Police’s use of ‘precision policing’ and related data analysis.
Specifically, I would like the following information in relation to ‘precision policing’:
1. What models are currently in use or development using ‘precision policing’ methods, in addition to models for ranking “dangerous men suspected of violence against women and girls”, knife crime stop and search, and a further model “to rank suspects on their entire criminal history over several years” as reported in July 2023?
2. For each model that is currently in use or development, please provide the following:
a. The names of the models or how they are referred to operationally;
b. What types of data are being used (or if in development, are being considered for use).
For example, PNC criminal history data, nationality data, ethnicity data, education data, employment data, immigration status data, relationship status data, financial data.
c. What types of operational action the outcomes or predictions of the model are (or will be) used for (e.g allocation of resources, patrols, monitoring, questioning, targeted interventions);
d. Copies of the Data Protection Impact Assessments and/or Equality Impact Assessments that have been carried out in relation to the model.
e. Copies of internal reviews or assessments relating to the models.
I have today decided to disclose some of the requested information. Some data has been withheld as it is exempt from disclosure and therefore this response serves as a Refusal Notice under Section 17 of the Freedom of Information Act 2000 (the Act) by virtue of the following exemptions:
Section 22 (1)(a) - Information Intended for Future Publication.
Reason for decision
The Act creates a statutory right of access to information held by public authorities. A public authority in receipt of a request must, if permitted, state under Section 1(a) of the Act, whether it holds the requested information and, if held, then communicate that information to the applicant under Section 1(b) of the Act.
The right of access to information is not without exception and is subject to a number of exemptions, which are designed to enable public authorities to withhold information that is unsuitable for release. Importantly, the Act is designed to place information into the public domain, that is, once access to information is granted to one person under the Act, it is then considered public information and must be communicated to any individual should a request be received.
The information you have requested. Question 2d:
Q2d - Copies of the Data Protection Impact Assessments and/or Equality Impact Assessments that have been carried out in relation to the model.
Is exempt by the virtue of the following Act:
Section 22 (1)(a) - Information Intended for Future Publication - Under Section 22 of the Act, Public Authorities are able to withhold information, if the information is held with a view to its publication.
The MPS is in the process of publishing DPIAs (Data Protection Impact Assessments) and EIAs (Environmental Impact Assessments) and to pause this process to prepare the information requested for disclosure under the Act, will interfere with the process already in hand to make this information accessible to the public along with other useful information.
Dealing with requests for information, which have already been identified as useful information for disclosure, helps the MPS to understand what is interesting to the public. The MPS had already intended to prepare information (Data Protection Impact Assessments and Equality Impact Assessments) in relation to the model.
Thereby promoting transparency and building positive communications with the public.
The information requested will be published in future publications via the MPS website: Publication date not specified.
The MPS is in the process of gathering information and to pause this process to prepare the information requested for disclosure under the Act, will interfere with the process already in hand to make this information accessible to the public along with other useful information.
I have determined that the disclosure of the above information would not be in the public interest at this time. We consider that the benefit that would result from the information being disclosed now, does not outweigh the considerations favouring non-disclosure, especially considering it will be available in the near future.
Disclosure
Q1 - What models are currently in use or development using ‘precision policing’ methods, in addition to models for ranking “dangerous men suspected of violence against women and girls”, knife crime, stop and search, and a further model “to rank suspects on their entire criminal history over several years” as reported in July 2023?
For the VAWG 100 project, we use the Cambridge Crime Harm Index (CCHI) to assign a score for each offence where an individual has been listed as a named suspect of a VAWG offence in the preceding three hundred and sixty five days. We then sum all of that individual’s scores, to provide us with their total CCHI score. We then rank all individuals according to the total CCHI. The individuals’ criminal history prior to the preceding three hundred and sixty five days does not impact this ranking.
The MPS is testing the feasibility of applying machine learning to predict the likelihood, amount, and seriousness of future VAWG offending within a relatively limited population of potential offenders who have a recent history of previous VAWG offending. The application of these techniques requires a significant deal of testing and this process is therefore simply at the early development stages at this point.
The Violence Harm Index is a tool used to identify the most harmful and violent offenders in London. It is our intention to publish information relating to the Violence Harm Index on the MPS website when we launch the VHI.
The method for selecting pilot locations for Precision Stop & Search is to use the CCHI scores for weapons crimes against personal victims, including stabbings and robberies, as aggregated in each of the 679 Wards over 365 days. Each Ward was ranked from 1 to 679 for the highest harm scores generated by reports of the eligible offence categories.
Dauntless+ is an initiative to identify, monitor and disrupt individuals who, in a domestic abuse setting, target and pose an ongoing and immediate risk to others though their offending behaviour. It uses the Cambridge Crime Harm index in tandem with recent, frequency, and gravity scoring.
Q2 - For each model that is currently in use or development, please provide the following:
a. The names of the models or how they are referred to operationally;
• VAWG 100
• Violence Harm Index
• Precision Stop & Search
• Dauntless+
B - What types of data are being used (or if in development, are being considered for use). For example, PNC criminal history data, nationality data, ethnicity data, education data, employment data, immigration status data, relationship status data, financial data.
The MPS use data from PNC and PND, as well as data from other policing systems.
C. What types of operational action the outcomes or predictions of the model are (or will be) used for (e.g. allocation of resources, patrols, monitoring, questioning, targeted interventions);
The models will influence allocation of police resources.
E. Copies of internal reviews or assessments relating to the models.
These documents do not currently exist.