Question to the Home Office:

To ask the Secretary of State for the Home Department, with reference to her Department's new story entitled PoliceAI to speed up investigations and fight crime, published on 10 June 2026, what mechanism exists to ensure that a tool shown by post-deployment monitoring to have suffered a substantial decline in accuracy is withdrawn; and who will be accountable if a Chief Constable chooses not to act on PoliceAI's advice.

To build and maintain public consent for the police use of AI, the police must test and deploy AI responsibly. That means deployments must be lawful, transparent, ethical, well governed and based on robust assessments of accuracy and bias. A core reason for setting up PoliceAI is to support Chief Constables in making informed decisions about the AI they use and to assist with responsible adoption.

The government is committed to transparency on police use of AI and PoliceAI will publish and maintain a registry of AI tools in use across policing. This is being developed in partnership with academics and is in testing with forces. We intend to launch it as soon as possible this year. The registry will include details on the steps taken by the force to deploy the AI responsibly, including ethical assessment and accountability arrangements.

On accountability, it is up to Chief Constables what AI they deploy and procure, accountable to their elected Police and Crime Commissioners (PCCs). PCCs and the public will be able to scrutinise decisions with the assistance of the registry, so there will be transparency if Chiefs ignore or do not make use of PoliceAI’s expertise. As a backstop, there are powers in Section 53 and 53a of the Police Act 1996 which the College of Policing and/or the Home Secretary can use to regulate police use of AI. Use of these is kept under review.