The UK government’s proposed , designed to flag individuals deemed “high risk” for future violence based on personal data like mental health history and addiction, marks a provocative new frontier.
Elsewhere, intends to use machine learning for crime prediction and real-time surveillance. In Canada, police forces in cities like Toronto and Vancouver use . And in some US cities, is paired with street surveillance to track suspects.
Advertisement
The promise of anticipating violence Minority Report-style is compelling. But any apparent utopia comes with substantial risks. Such technologies can easily be misapplied, especially when human complexity is reduced to data points. This shift towards prediction echoes the logic behind , a pillar of the UK’s counterterrorism strategy, which drew , turning teachers, doctors and social workers into de facto informants, and eroding public trust in public institutions.
We now risk repeating those failures – this time with machine learning’s illusion of neutrality. Algorithms are only as fair as the data they are built on. In UK policing, . Embedding those biases into code doesn’t fix the problem; it hardwires injustice into decision-making.
that AI could revive and scale the same kind of profiling that made Prevent so controversial. Mohammad Al-Issa, secretary-general of the League, has AI could generate to fuel division and extremism.
His concerns, grounded in efforts to counter and the digital recruitment tactics of groups like the Islamic State, reflect a growing consensus: . If misused, it can deepen mistrust and undermine the social fabric that prevention policies are meant to protect.
The idea of early intervention isn’t inherently wrong. AI could support a more humane, proactive approach to justice. Tools to analyse behaviour patterns might help identify people before they resort to violence, connecting them with mental health support.
Police and probation officers could more efficiently allocate resources to the most urgent cases. Algorithms might also help reduce some forms of subjective bias, offering greater consistency in how cases are assessed.
But we must be honest about the and the need for these systems to have rigorous oversight. False positives remain a critical risk. An individual could face stigma, scrutiny or police intervention without ever having committed a crime.
The danger of mission creep is also real. Just as with Prevent, an AI tool built for serious crime prevention could gradually expand to monitor a wider range of people or beliefs – especially those already over-policed and low on trust in the justice system.
As chair of the advisory board to the UK’s All-Party Parliamentary Group on the Metaverse and Web 3.0, I’ve seen how quickly technology outpaces ethics. Yet I remain hopeful that AI can reduce harm – not just respond to it – if we get the design, deployment and oversight right. If we have learned anything from past missteps, it’s that true security begins with trust – not surveillance. Prevention means investing in people: in mental health, education and communities. AI can help scale those solutions. But we must always ask not just what it can do, but whether it should.
Professor Yu Xiong is associate vice president of the University of Surrey, UK
