Policing in the Age of AI
How artificial intelligence is redefining policing in London and the challenge of preserving policing by consent.
As a new technological revolution takes hold, we are forced to confront the question: just because we have the capability, should we use it? Artificial intelligence has already reshaped how we communicate, how we work and how we fight wars. It is now reshaping how we are policed. With the government committing £75 million over three years to launch PoliceAI, Britain has placed a substantial bet on artificial intelligence as the future of law enforcement. As Neil Basu, the former head of Counter Terrorism Policing, acknowledged, “there is a lot of concern about AI but the truth is it is here, and it’s here to stay”. The question facing us today is therefore not whether AI belongs in policing, because it has already taken root, but whether we can embrace its benefits without compromising the principle that has underwritten British policing since 1829: policing by consent.
Much of the policy is aimed not at replacing officers but at eliminating bureaucracy. AI is already transcribing 999 and 101 calls, analysing CCTV footage and processing vast quantities of digital evidence in a fraction of the time it would take a human investigator. Modern policing suffers less from a lack of information than from an inability to connect it. Intelligence sits scattered across databases, body-worn cameras, witness statements and crime reports. AI’s greatest strength lies in bringing these fragments together, revealing patterns that would otherwise remain hidden. If successful, it could free officers to spend less time behind desks and more time in the communities they serve.
Live Facial Recognition (LFR) illustrates both the opportunities and concerns surrounding AI in policing. Critics argue that facial recognition represents a move towards mass surveillance, yet the Metropolitan Police’s system is not designed to create a database of every member of the public. Instead, it compares images against a specific watchlist of individuals wanted by the police or subject to particular policing objectives. The technology has also demonstrated operational value, with LFR reportedly almost three times more likely to lead to an arrest than a conventional stop and search. It certainly has the potential to make policing more targeted rather than more intrusive.
However, while the Tony Blair Institute does have a point in their argument that the potential of AI has been overshadowed by public debate disproportionately focused on its risks, concerns around artificial intelligence cannot simply be dismissed. If past policing practices have disproportionately affected certain demographics or neighbourhoods, AI systems risk reproducing those patterns under the appearance of objective mathematics. Algorithms are only as neutral as the data and decisions that shape them. Predictive policing tools, in particular, raise concerns about creating a feedback loop where historical patterns of enforcement influence future policing priorities.
These concerns become even more significant as Britain becomes increasingly reliant on private technology companies. The Metropolitan Police’s proposed partnership with American technology firm Palantir has raised questions about data sovereignty, civil liberties and democratic accountability. Palantir already provides technology to organisations including the Ministry of Defence and the NHS, but its involvement in policing has proven more controversial. The Mayor’s Office for Policing and Crime previously raised concerns about whether Palantir’s values aligned with those of London, while Palantir has since taken legal action seeking to overturn the decision not to approve the contract. Commissioner Sir Mark Rowley has argued that the £25.3 million agreement is necessary. His defence of AI rests on a practical argument: Parliament cannot legislate for every technological development because innovation moves quickly while legislation moves slowly, and there is undoubtedly merit to this claim. If every new capability required fresh primary legislation before it could be deployed, policing would inevitably fall behind the criminals it seeks to catch. Effective oversight therefore cannot rely solely on new legislation for every technological development. It must ensure that police forces can explain how AI systems are used, what safeguards are in place and who is accountable for their decisions.
Britain also faces a far larger dilemma. If the country wishes to reduce its dependence on American technology firms, the answer cannot be to ask the police to use inferior tools. In an age where criminals exploit AI to automate fraud, generate deepfakes and carry out increasingly sophisticated cybercrime, expecting the police to fight 21st-century threats with outdated technology is hardly a sustainable strategy. The responsibility for developing competitive British and European AI companies lies with government, industrial policy and the wider economy, not the Metropolitan Police. Officers cannot conjure domestic suppliers into existence, and it would be unreasonable to expect them to police London less effectively simply because better alternatives are currently developed overseas.
The future of AI in British policing therefore lies heavily on the standards surrounding its use not just on the power of the technology. Policing by consent remains fundamental because the authority of the police is not derived solely from their legal powers, but also from public confidence that those powers are exercised fairly and responsibly. Britain should not choose between innovation and consent but must instead ensure that innovation serves consent. Thus, Sir Robert Peel’s principles remain as relevant today as they were in 1829: the police must ensure that the use of even the most transformative technologies maintains the trust of the communities they serve.
By Rose Sarkar
