Artificial Intelligence
The latest (10 July 2026) in the Probation Inspectorate’s Academic Insights series explores the current and potential uses of Artificial Intelligence (AI) within the Probation Service. Produced by Elizabeth McClory-Tiarks, Claire Paterson-Young, Michael Maher and Michele Sevegnani, the report highlights the opportunities presented and the challenges raised for practice, decision-making, and service delivery.
It is clear that the direction of travel is one of increasing experimentation, with AI-driven tools having been proposed in the areas of information retrieval, transcription and summarisation, risk assessment, sentence planning, resource allocation, compliance monitoring, and early identification of reoffending risks.
Current uses
There is currently limited use of AI within the Probation Service, but nDelius, the probation case management system, is one example where AI has been integrated and is intended to reduce the administrative workload of probation staff.
As much of the information needed by practitioners exists in large volumes of unstructured text, the platform has integrated a LLM-based semantic search function which helps practitioners locate semantically similar case notes based on the user’s search criteria, even when different wording or misspellings are used.
Many practitioners also use Justice Transcribe, an in-house AI transcription and meeting summarisation tool to expedite the process of recording supervision notes.
Assessing Risks, Needs and Strengths (ARNS), the proposed replacement for OASys will produce risk assessments, risk management plans, and pre- and post-sentence plans, drawing on actuarial reoffending data. It also introduces automated features, such as text message appointment reminders to improve compliance and tools to streamline the administrative processes to increase efficiency within the Probation Service.
The authors say they understand that ARNS will also provide the digital infrastructure for the further integration of AI.
Potential future use
The authors note that the AI Action Plan for Justice sets out a vision for the possible beneficial uses of AI, including the personalisation of rehabilitation plans and AI-enabled predictive tools for risk assessments.
They note that the Independent Sentencing Review suggests that an AI enabled risk assessment tool could be combined with facial recognition to track an individual’s movements in public spaces. This is with a view to these tools enabling probation officers to detect warning signs, adjust supervision levels and intervene when needed.
The Sentencing Review also proposes the use of AI tools to personalise probation services to people’s differing needs, risk levels and in different situations: “mobile-based rehabilitation apps and tools could be used to enable an offender to connect with the Probation Service and access information more easily, all in one place on their phone”.
Opportunities and risks
The authors discuss the potential benefits of AI helping probation practitioners to reduce the (considerable) administrative burdens of their job but also caution that the introduction of AI into probation raises ethical, legal, and professional challenges. The authors highlight three key risk areas:
- bias, discrimination and privacy
- lack of transparency and overreliance on technology
- due process and accountability.
The authors recommend a responsible AI framework like the one reproduced below.
Conclusion
The authors conclude by saying that AI will be welcomed if it can reduce administrative work and enable probation practitioners to spend more time working directly with people on probation. At the same time, they strongly recommend a probation-specific AI framework which:
“help to ensure the best quality decision-making in a high-stakes environment, maximising benefits and minimising risks.”
Thanks to Andrea De Santis for kind permission to use the header image in this post which was previously published on Unsplash.
