AI and Technology: Whose assumptions are being automated?

Technology can help a stretched system manage information and reduce duplication. But in criminal justice, a tool is never just a tool. It can influence who is marked as risky, what support is offered and whose account is believed. The Oxford Institute of Technology and Justice’s mapping of the UK shows AI being used across the criminal process, from policing and evidence review to assessment and case management.

For Black and Global Majority communities, that raises a hard question. What happens when systems built on historic data and unequal decisions are taught to make recommendations about the future? Bias does not disappear because it has been placed inside a spreadsheet or a model. It can become harder to see, challenge and reverse.

The Ministry of Justice says AI in justice should protect rights, work within the law and support, not replace, human judgement. Good. But those principles need to show up in practice. Before a justice agency buys or rolls out a new system, practitioners and communities should be able to ask: What data trained it? What evidence shows it works fairly? Who can challenge an automated recommendation? What racial-disparity testing has been published? And who has the authority to stop it when harm appears?

The principle is straightforward: technology should strengthen community power, not automate old inequalities. Last year BM4C began incubating a special purpose vehicle called Black AI Futures. It was launched in June 2025 and is our contribution to being the change, rather than merely lamenting the risks.

One organisation, Breakthrough Social Enterprise, is leading the way. Read the accompanying article by Pratik Doshi of Breakthrough Social Enterprise.

Sources and further reading

• Oxford Institute of Technology and Justice: UK AI Justice Atlas
• Ministry of Justice: AI Action Plan for Justice