British police admit bias in their AI, but promise to tackle it

When a high-ranking police official openly states that the technology the state is about to deploy en masse contains systemic bias, this deserves close attention. This is precisely what happened in the United Kingdom, where Alex Murray, head of the AI division at the National Crime Agency (NCA), in an exclusive Guardian interview acknowledged the obvious: artificial intelligence in law enforcement will be biased. But he immediately promised that this would be addressed.

The context of this statement is no less important than its content. The Labour government of the United Kingdom has embarked on a dramatic expansion of AI use in the police of England and Wales. This is not about local experiments, but about systemic transformation: police leadership is convinced that without artificial intelligence, law enforcement will simply be unable to keep pace with the evolution of crime. To realize these ambitions, a specialized police AI center is being created with a budget of 115 million pounds sterling—roughly 13.5 billion rubles at the current rate. A sum that speaks to the seriousness of intentions.

Murray himself hastened to reassure the public with a phrase that has already become a meme in British media: "This is not RoboCop." According to him, the focus is primarily on improving efficiency in complex investigations—processing large datasets, identifying patterns, accelerating analytical work. It sounds reasonable and even harmless. But the devil, as always, is in the details.

The problem of AI bias in law enforcement is not an abstract theoretical threat. Global experience has already accumulated enough alarming examples. In the United States, predictive policing analytics systems have repeatedly demonstrated racial bias, directing disproportionately more resources to neighborhoods where minorities live. Facial recognition systems showed significantly higher error rates when identifying people with dark skin. Algorithms for assessing recidivism risk assigned higher scores to representatives of certain ethnic groups under otherwise equal conditions. All of this is not hypothesis—these are documented cases.

The fact that Murray openly acknowledges the presence of bias can be interpreted in two ways. On one hand, it is a manifestation of intellectual honesty, rare for government officials promoting costly technology projects. Usually at this stage, people speak of "algorithmic neutrality" and "data objectivity." On the other hand, the formula "we know there is a problem, and we will work on it" is a classic expectation management technique that allows technology to be implemented now and problems to be addressed later. A promise to "limit injustice" is not the same as a promise to eliminate it.

For broader context, it is important to understand that the United Kingdom is following a global trend. Law enforcement agencies worldwide are increasing their use of AI, and the question is no longer whether police will apply these technologies, but on what terms. The European Union in its AI Act has established strict restrictions on the use of real-time biometric identification systems. The United Kingdom after Brexit is not bound by these rules and, apparently, is choosing a more liberal approach—with an emphasis on self-regulation and internal control.

The architecture of oversight itself deserves special attention. The creation of a specialized center for 115 million pounds is an attempt to centralize development and supervision, which is theoretically better than chaotic AI implementation by individual police departments. However, a key question remains unanswered: who will monitor the monitors? If the police conduct the audit of bias themselves, it is as if we are asking a fox to guard the henhouse. Independent external oversight, algorithm transparency, mechanisms for appealing decisions—all of this remains in the realm of good intentions for now.

For the Russian audience, this story is important as a reference point. Domestic law enforcement agencies are also actively implementing facial recognition technologies and predictive analytics, but there is almost no public discussion about the bias of these systems in Russia. The British experience shows that even in a country with developed institutions of civil control and independent media, the problem of AI bias in police remains unresolved. Acknowledging the problem is a necessary first step, but without concrete oversight mechanisms, it risks remaining a beautiful gesture followed by the quiet implementation of imperfect technologies on a nationwide scale.