Where this differs from ordinary deployment
Facial recognition, drones and crowd analytics are increasingly used at large public gatherings, and the way they are used at a music festival or a carnival is not always easy to separate cleanly from how the same infrastructure could be turned toward a protest or a political demonstration. That distinction matters legally as well as practically. Routine criminal policing engages rights to privacy, but a demonstration or march additionally engages the right to freedom of assembly and expression, protected under the European Convention on Human Rights and reflected in domestic law, which means surveillance technology deployed at a protest is judged, or ought to be judged, against a different and in some respects higher legal bar than the same technology deployed at an ordinary crime hotspot.
Notting Hill Carnival provides the clearest long running UK case study, precisely because it sits at the boundary between the two categories: a celebration rather than a protest, but one with a fraught history between organisers and the Metropolitan Police, at which live facial recognition has been trialled, criticised, withdrawn and expanded again repeatedly since 2016.
The Notting Hill Carnival case study
The Met first tested facial recognition at the carnival in 2016. Early results were poor enough that campaign groups had a genuine evidential basis for objection: one deployment produced no arrests and only two correct matches against the watchlist being searched, prompting the Met to withdraw the technology from at least one subsequent year's event following pressure from Big Brother Watch, Liberty and Privacy International, who argued that testing a technology at the same event year after year could not reasonably be described as a trial. Deployment resumed and has expanded substantially since. At the 2026 carnival, marking the event's 60th anniversary, live facial recognition generated 214 alerts over the two day event, contributing to 121 of the 636 total arrests, and allowed officers to identify and speak to 86 registered sex offenders in the crowd, seven of whom were arrested for breaching their conditions. The Met has credited the technology with a specific and repeatable public safety function at the event, distinct from its more contested use scanning general crowds for unspecified persons of interest.
The oversight that exists around this deployment is mainly political rather than judicial: the London Assembly's Police and Crime Committee holds annual question and answer sessions with the Deputy Mayor for Policing and Crime specifically examining the carnival operation, including facial recognition's role in it, providing a recurring, if not legally binding, form of public scrutiny that most other UK facial recognition deployments do not receive on anything like the same regular schedule.
When the same infrastructure is proposed for protest specifically
The clearest illustration of the underlying concern surfaced in London Assembly questioning of the Mayor in 2026, when a member of the Assembly asked directly whether live facial recognition and drone surveillance, both justified publicly by their use at events such as the carnival, could also be pointed at political demonstrations she described critically as hate marches. The exchange was notable less for the answer given, that deployment decisions of that kind are treated as an operational matter for the police rather than the Mayor personally, than for what the question itself demonstrated: capability built and publicly justified for one purpose, monitoring a large crowd for known offenders at a specific recurring event, creates an available tool that political actors will then propose redirecting toward monitoring political assembly and dissent, a substantially different application with a substantially different legal and democratic justification required.
Separately, the Met's plans to install permanent live facial recognition infrastructure on Oxford Street, expected before the end of 2026, mark a further shift, from event specific temporary deployment toward permanent scanning of one of the country's busiest public streets, which by its nature will capture anyone passing through the area for any reason, including anyone travelling to or from a protest that happens to route through or near it, without that being the stated purpose of the deployment at all.
What is not settled
There is no separate UK legal framework specifically governing the use of facial recognition, drones or crowd analytics at protests and demonstrations, as distinct from the general facial recognition law addressed elsewhere on this site, meaning the same unsettled legal position that applies to routine deployment applies here too, without additional protection specific to assembly rights beyond what the Human Rights Act already provides in principle. Nor is there a transparent, published account of how frequently technology justified by its use at events such as carnivals or festivals has in practice also been deployed at explicitly political gatherings, a gap that makes it difficult to assess, from public information alone, how far the concern raised in the London Assembly exchange described above reflects an existing practice rather than a hypothetical one.
Follow the coverage
PoliceAI News tracks every protest policing story from across the English-speaking world as it breaks: new deployments, court rulings, parliamentary debates, academic research, and community responses. The feed refreshes every 30 minutes.
View Live Protest Policing StoriesYou can also browse every protest policing story in the feed, or explore every subject PoliceAI News covers.