Guidance

Preventative Analytics for Youth Justice Advisory Panel: Terms of Reference

Published 2 March 2026

Applies to England and Wales

Purpose

The principal aim of the youth justice system, as set out in the Crime and Disorder Act 1998, is to prevent offending by children. This places a clear duty on services to act early to support children to avoid entering or becoming further entrenched in the youth justice system.

Advances in machine learning, artificial intelligence, and advanced analytics have the potential to strengthen this preventative aim by enabling earlier identification of need, more accurate understanding of risk, and better targeting of evidence-led support. Used appropriately, these tools may help services intervene sooner, manage risk more effectively, and deploy scarce resources where they can have the greatest impact. In doing so, they offer opportunities not only to improve outcomes for children, but also to enhance public protection and contribute to the prevention of crime by reducing the likelihood of escalation into more serious or persistent offending. The potential to harness these technologies in a way that supports both prevention and public safety therefore warrants careful consideration.

Implementing advanced analytical techniques in youth justice is an area of high sensitivity, public interest, and ethical complexity. These approaches must be applied in alignment with the best long-term interests of the child, build on their individual strengths and capabilities, encourage active participation, and promote prevention, diversion and minimal intervention so that contact with the justice system, and the stigma that can accompany it, is kept to a minimum. Children within the youth justice system are a uniquely vulnerable cohort, and therefore any use of modelling or algorithms must be:

• aligned with the best long-term interests of the child
• ethical (and in particular, aligned with the Ministry of Justice’s https://www.gov.uk/government/publications/ministry-of-justice-ai-and-data-science-ethics-framework)
• legal
• technically robust
• operationally workable

The Preventative Analytics for Youth Justice Advisory Panel (PAYJAP) will suggest and review proposals for utilising machine learning, artificial intelligence, and advanced analytics, providing an additional layer of accountability and scrutiny, ensuring that any data science or algorithmic development is safe and appropriate, well governed and in line with the above considerations.

Scope

PAYJAP will consider:

• the suitability of advanced analytics, artificial intelligence, and machine learning models for providing valid, accurate and meaningful insights
• potential data processing approaches used to construct these tools
• whether proposed solutions align with the best long-term interests of the child
• ethical questions around fairness, explainability, proportionality, and consent.
• Operational implications and frontline usability
• relevant legal considerations including data protection, liability and safeguarding
• alignment with Ministry of Justice (MOJ), Youth Justice Board, and cross-government standards for algorithmic governance

Governance

The Panel is advisory, not a formal decision-making body.

Meetings

Minister Richards will chair the meeting and Ministry of Justice Service Transformation Group will provide the Secretariat.

The PAYJAP will meet quarterly. Where possible, the agenda and relevant papers will be circulated one week prior to the scheduled meeting.

Membership

PAYJAP will consist of internal formal members and named independent advisors who will attend each meeting.
Core membership as follows:

• Chair: Jake Richards, Parliamentary Under-Secretary of State for Justice
• Chief Data Officer, Ministry of Justice
• Deputy Director, Youth Justice Policy Unit, Ministry of Justice
• Chief Executive Officer, Youth Justice Board

Professor Mark Mon-Williams, Chair in Cognitive Psychology at the University of Leeds, has agreed to serve as an independent advisor and aid in establishing panel. We are seeking further external independent advisors to join the panel. We intend to recruit members with expertise in one or more of the following:

• youth (re)offending
• machine learning or algorithmic decision support. We would particularly welcome individuals with expertise in applying machine learning in justice or public policy settings
• operational workings of the youth justice system (e.g. gained by experience in a youth justice service, charity, front-line child psychology, or the police)
• legal frameworks for use of data and analytics with children
• ethical use of data and algorithms

To note, the independent advisors will not be remunerated.

Review arrangements

The Secretariat will formally review these terms of reference periodically to ensure the arrangements are effective. Any amendments will be discussed with the panel members and agreed by the Chair.

Confidentiality

MOJ may share confidential information, data and ideas that are at an early stage of development to encourage free and open discussions between the panel members. Information which is not already in the public domain must not be shared publicly or with third parties. NDAs may be arranged when deemed necessary.