AI in Probation and the Future of Public Service Reform

Amid Austerity, a Glimmer of Reform: What the Spring Statement Means for Probation and AI.

The Spring Statement 2025 didn’t deliver many surprises. As the economy wavers and borrowing costs rise, Reeves reinforced Labour’s fiscal discipline, which includes controversial welfare cuts and a new wave of civil service reform. But buried in the detail was something far more forward-looking: a £3.25bn “transformation fund” to modernise public services, including investment in AI tools for the Probation Service.
Modular Data welcomes the ambition as a data consultancy working in both prisons and probation. But if this is to be more than another round of technocratic reform, then we need to focus on the foundations if it is to help reduce reoffending and support frontline workers.

The Spring Statement 2025 confirmed what many in the public sector have been anticipating: the government is moving forward with its ambitions to transform public services through technology. Against the backdrop of budget constraints, rising demand and system strain, the Chancellor’s announcement of a new £3.25bn transformation fund marked a significant moment in the direction of reform. The first tranche of this fund includes investment in AI tools for the probation service—an area under growing pressure as the justice system teeters near gridlock.
For those of us working at the intersection of data, AI, and public service delivery, the announcement is both a challenge and an opportunity. Modular Data has been supporting transformation across the justice system for several years. We have worked closely with prison colleagues and are now collaborating with the probation service to improve the systems that support rehabilitation, reduce reoffending, and protect the public.

The current state of probation and prisons demands urgent and focused action. The Public Accounts Committee has described the prison estate as being on the brink of collapse. Occupancy is at 99.7% across the adult male estate, with thousands of prisoners sharing cells that were never designed for more than one person. Violence is rising. Staff are overstretched. Access to education, health assessments and drug rehabilitation is increasingly compromised. Prison should be a place for rehabilitation and reform. Instead, it is becoming a holding system driven by risk and reactivity.

In this context, the probation service carries a growing weight of responsibility. As prison capacity evaporates, effective probation becomes the primary line of defence in reducing reoffending and managing the reintegration of individuals back into society. It is encouraging that the Justice Secretary has set out a clear ambition for the probation workforce: recruit more staff, strengthen capability, improve retention, and embed wellbeing into operational culture. This aligns with a larger vision of public safety built on trust, data and modern technology.
However, improving probation is not simply a question of headcount. The volume and complexity of caseloads continue to rise, and the administrative burden is high. Caseworkers must balance risk assessment, rehabilitation planning, court reporting, community engagement, and safeguarding duties. This is where the role of AI and data becomes most promising.

The idea that AI can shoulder administrative work and allow probation staff to focus on human relationships and critical decisions is appealing. Yet technology cannot be deployed in isolation. Public sector AI must be designed with ethics, transparency and service delivery. In probation, there is little room for error. Every insight, prediction and recommendation must be explainable. Every dataset must be governed. The need for clarity and trust in AI is not optional—it is fundamental.

There is a danger, however, in jumping too quickly from aspiration to automation. The Spring Statement aims to cut civil service costs by 15% and deliver £3.5bn of savings in government operations by the end of the decade. The temptation to see AI as a cost-saving mechanism first rather than a tool for public value remains strong. This framing can undermine public trust and lead to ineffective implementation. What matters is building capability and understanding what problems we are solving.

In probation, data can already support continuous improvement. By focusing on essential analytical questions, we can enable smarter decisions without rushing to replace human oversight. What factors shape the customer environment, and how can we respond to them in real-time? How effective are our interventions, and what outcomes are we achieving? Are we capturing accurate information the first time and reducing the need to ask the same questions repeatedly? Where is our cost being consumed, and how can we shift resources towards impact?

There are practical steps we can take. We can design systems that make data accessible to frontline teams without compromising security or privacy. We can build analytics pipelines that reduce the time it takes to gain insight. We can create shared service delivery models across probation, prisons, housing and health so that data follows the person, not the department. We can identify and address the causes of demand for failure through better data quality and shared accountability.

To do this effectively, we must move beyond dashboards and reports. We must develop data products—governed assets that combine information, logic, and policy in reusable, scalable and safe ways. These products must be open, explainable and composable. They should support the probation workforce by surfacing the proper insight at the right time. They must integrate with the broader justice system without creating new silos.

The move towards AI in government must be grounded in this approach. Before models can make decisions, we need to ensure the foundations are in place. Governance must be embedded from the start. Access and control must be granular, not generalised. We must build trust with the public and public servants, ensuring every system can be understood and interrogated.

The transformation fund announced in the Spring Statement is welcome. So is the focus on probation technology. But the conversation must now move to specifics. What capabilities are we building? What problems are we solving? How will we measure success? These questions require answers rooted in delivery, not hype.

As Modular Data supports the next phase of justice reform, our focus remains the same. We aim to build open, ethical, and responsible data and AI systems that help public servants do their jobs better. The future of public service transformation depends on empowering people with trustworthy technology. Probation is no exception. It deserves systems that reflect its complexity and support its mission.

The choices we make now will shape the role of AI in government for years to come. We must get them right.