Counterterrorism: AI and Facial Recognition

Baroness Blower Excerpts

Wednesday 1st July 2026 Lords Chamber

[Lord Hanson of Flint (Lab)]

At the moment, as the noble Lord knows, the Government’s policy objective is to reduce the number of police forces. We have asked the noble Lord, Lord Hogan-Howe, to look at that. Ministers have not yet had his final report with his proposals for the size of police forces. In the police reform Bill, we are trying to establish national responsibilities on crime, counterterrorism, training, and procurement. As part of our consultation, we are looking at what we need to do on a framework for facial recognition technology. As the noble Lord rightly said, we have put in £140 million, including £11 million on 40 new live facial recognition units, and we have looked at retrospective facial recognition. We are trying, in a very long-term plan, to get national capability, better resource and better efficiency for taxpayers, while at the same time recognising, as I know he does, that facial recognition is an extremely important tool for identifying known criminals, people who have broken licence terms and people who are missing, and for intelligence-led policing that can ensure that we know exactly who we are looking for and why.

[Baroness Blower (Lab)]

My Lords, this is clearly a very useful technology, but given that it has very wide application, can my noble friend say anything about how we monitor the level of possible misrecognition by these systems?

[Lord Hanson of Flint (Lab)]

That is an absolutely vital point, as it is with many forms of technology. The old identity parade misidentified some people. ANPR—number plate recognition—occasionally looks at information that does not identify the right number plate. This is technology that is used to support the police. I reassure my noble friend that at the end of it there is always a set of human eyes. Humans determine whether to put a case to the CPS for prosecution and whether the information is correct. But AI, including AI used for facial recognition, means that we are saving the police hours, to the extent that we can potentially deploy an extra 3,000 police officer hours for other purposes. That is because AI can sift out, examine, look at characteristics and determine one, five or even 10 suspects who we might need to have further eyes on. It is important to look at the developments. Ultimately, it is under operational police and political control.