
People attend demonstrations for many reasons: to oppose government policy, support a cause, express anger, remember a tragedy or demand change. In a democracy, the ability to gather in public and be heard is a fundamental freedom. Facial recognition introduces a new complication. It can allow authorities to identify individuals in a crowd, potentially linking a person’s presence at a protest to a name, a police database or an earlier investigation.
- How police facial recognition can be used
- The police case: identifying serious offenders
- The risk of a chilling effect
- What the courts have said
- The government’s proposed legal framework
- What safeguards would make a difference?
- Can protest surveillance ever be justified?
- Security and democratic freedom
- Sources and further reading
Police argue that the technology can help locate people wanted for serious offences and protect the public at large events. Civil liberties groups warn that scanning people who are exercising lawful rights can undermine privacy and discourage participation. The dispute is not simply about whether the technology works. It is about when the state should be allowed to identify people who have not individually attracted suspicion.
How police facial recognition can be used
Live facial recognition analyses images from cameras as people pass through a monitored area. The system detects faces and compares them with a watchlist. If it identifies a possible match, an alert is sent to an operator or officer, who is expected to assess the result before deciding whether to intervene.
Retrospective facial recognition works differently. Investigators may compare an image from CCTV or another recording with photographs held in a database after an incident has occurred. An operator-initiated search can involve an officer submitting a particular image for comparison. These methods differ in timing and scale, but each can create questions about the accuracy, purpose and retention of biometric information.
At a protest, live scanning can involve people who are not suspected of any offence. The system may process thousands of faces to identify a small number of people on a watchlist. Even if the police say that the watchlist is limited, the wider crowd has still been subjected to biometric analysis.
The police case: identifying serious offenders
Police forces say live facial recognition can help identify people wanted for serious violence, sexual offences and other crimes, including individuals who may otherwise be difficult to locate. The Metropolitan Police has reported substantial numbers of arrests linked to its deployments in London. In May 2026, Reuters reported the Met’s claim that the technology had helped arrest about 2,500 wanted people since early 2024.
During operations described in the same report, cameras scanned large numbers of passers-by against a watchlist. Police argued that the system generated very few false alerts relative to the number of faces processed and that officers made decisions only after checking a possible match. Civil liberties campaigners disputed whether such a broad intrusion was justified, especially where the system was used near demonstrations.
These figures are important but should be understood as reported police data, not an independent finding that every deployment is necessary or proportionate. Arrests do not by themselves establish guilt, and the number of alerts must be considered alongside the accuracy of matches, the consequences of mistakes and the people whose faces are scanned without ever receiving an alert.
The risk of a chilling effect
A person may decide not to attend a demonstration if they believe their identity will be recorded and linked to their political activity. That concern can affect not only people who fear arrest, but also journalists, campaigners, trade union members, vulnerable witnesses and people who simply wish to keep their lawful activities private.
This is often described as a chilling effect. It does not require the police to arrest or even approach anyone. The possibility of being identified may be enough to change behaviour. If people feel that attendance at a public meeting or protest could create a permanent record, they may become less willing to participate in civic life.
The risk is especially serious when surveillance is used without a clear connection to a specific threat or criminal investigation. A peaceful demonstration should not automatically be treated as a security problem merely because it attracts a large crowd or expresses unpopular views.
What the courts have said
In Glukhin v Russia, decided in 2023, the European Court of Human Rights considered the use of facial recognition to identify and locate a man who had taken part in a peaceful one-person protest in Moscow. The Court found violations of his rights, including the right to respect for private life and freedom of expression. The case is a significant European human-rights judgment on the use of facial recognition in the context of peaceful protest.
In Britain, a 2020 Court of Appeal judgment concerning South Wales Police’s use of automated facial recognition found the particular policy and deployment unlawful. The court identified deficiencies in the legal framework and in the way the police had exercised discretion. The judgment did not establish that every possible use of facial recognition is unlawful, but it set out important requirements concerning safeguards and the exercise of police powers.
More recently, in April 2026, the High Court rejected a challenge to the Metropolitan Police’s current live facial recognition policy. Reuters reported that the court considered the policy sufficiently clear and its safeguards adequate in the circumstances of that case. That ruling is part of the developing legal picture, not a blanket approval of every deployment or every use at a protest.
The government’s proposed legal framework
In November 2026, the government published its response to a consultation on a new legal framework for law-enforcement use of biometrics and facial recognition. The response records concerns raised by respondents about privacy, freedom of expression and peaceful assembly, including calls for restrictions on use at demonstrations.
The government acknowledged that these technologies can interfere with rights and said it intends to establish clearer legal powers and obligations, supported by operational requirements and a statutory code of practice. It says use should be demonstrably justified, necessary and proportionate. The response is a policy step towards a new framework; it should not be confused with a complete set of new laws already in force.
What safeguards would make a difference?
Any deployment should have a clearly defined operational purpose and a documented explanation of why facial recognition is needed instead of less intrusive methods. The seriousness and likelihood of the threat should matter. So should the location, duration, size of the watchlist and the likely effect on people who are not suspected of wrongdoing.
Independent authorisation can provide an additional check before especially intrusive deployments. Watchlists should be limited to people who meet clear criteria, and entries should be accurate, current and subject to review. Police should publish information about where and why systems are used, how many people are scanned, how often alerts are confirmed and what happens to data relating to non-matches.
There must also be meaningful ways to challenge an incorrect match or an inappropriate watchlist entry. Human review is essential, but officers need training and should not treat an algorithmic alert as proof of identity or guilt. Retention periods, access controls and audit trails should be clear and enforceable.
Can protest surveillance ever be justified?
There are difficult cases. If credible intelligence indicates that a particular individual is planning a serious violent offence at an event, a targeted investigation may be justified under the law. Facial recognition might form one part of that response if strict conditions are met. But this is very different from routinely scanning everyone attending a peaceful march because the technology is available.
The distinction between targeted investigation and indiscriminate scanning is central. A public event does not remove a person’s rights, and the fact that a face is visible in a public place does not mean that every form of automated identification is proportionate.
Security and democratic freedom
Supporters of facial recognition see it as another investigative tool that can help police find dangerous suspects. Opponents fear that normalising biometric checks in public spaces could gradually change the relationship between citizens and the state. Both positions raise questions that deserve evidence rather than slogans.
A democratic society needs effective policing, but it also needs people to feel able to challenge authority, report wrongdoing and assemble peacefully without unnecessary surveillance. The legitimacy of facial recognition will depend not only on arrests or technical accuracy, but on the quality of the rules, the transparency of deployments and the willingness of authorities to accept limits.
Sources and further reading
- Home Office: Government response to consultation on a new legal framework (November 2026)
- Reuters: On London’s streets, facial recognition tests the balance between security and liberty (22 May 2026)
- Reuters: UK court rejects challenge to London police’s use of live facial recognition (21 April 2026)
- European Court of Human Rights: Glukhin v. Russia (2023)
- Court of Appeal: R (Bridges) v Chief Constable of South Wales Police (2020)
- Equality and Human Rights Commission: Policing and human rights
