Restore trust in police and end disproportionate policing tactics
Green · what the evidence says
An independent, source-checked look at Green’s policy “Restore trust in police and end disproportionate policing tactics” — what it would actually do across the things that affect your life. Every claim below quotes the source behind it. How this works.
Personal liberty & free speech — Helps
moderate · moderate confidence
Ending routine stop and search and facial recognition technology would directly reduce surveillance and state coercion of the public, improving personal liberty — especially for Black communities who face these powers most. The main caveat is that the policy uses the word 'routine', leaving scope for continued use, and diversity assessments on officers are an internal workplace measure with unclear liberty implications.
The evidence
- The policy commits to ending the routine use of stop and search and facial recognition software. — greenparty.org.uk (manifesto) — “End the routine use of stop and search and facial recognition software”
- Black people face stop and search at a rate 3.7 times higher than white people, representing a major disproportionality in coercive police contact. — commonslibrary.parliament.uk (government) — “Black or Black British people were searched at a rate 3.7 times higher than those from a White ethnic group across England and Wales”
- This racial disproportionality in stop and search has persisted for at least 20 years. — libertyhumanrights.org.uk (media) — “This disparity has been consistently high for 20 years”
- The Court of Appeal ruled in 2020 that South Wales Police's use of live facial recognition breached privacy, data protection, and equality laws. — emergencyservicestimes.com (media) — “The Court of Appeal ruled in 2020 that South Wales Police's use of live facial recognition (LFR) breached privacy, data protection, and equality laws”
- FRT systems exhibit documented racial and gender bias, with Black women subject to the highest false positive identification rates. — post.parliament.uk (government) — “Tests on an FRT used by UK police in 2025 found Black women were subject to the highest percentage (9.9%) of false positive identifications at a 0.8 threshold”
- Reducing stop and search disproportionality is expected to rebuild trust with communities that current practices have further marginalised. — libertyhumanrights.org.uk (media) — “Reduced disproportionality is expected to rebuild trust with communities, particularly Black communities, which current practices have "further marginalised"”
Biggest unknown: Whether 'ending routine use' means a near-total ban or merely a shift to more targeted use — the actual liberty gain depends heavily on how strictly this is enforced in practice.
Our reading: O10 is directly concerned with freedom from state coercion, surveillance, and bodily autonomy. Both instruments targeted by this policy — routine stop and search and facial recognition technology — are paradigm cases of state coercion and surveillance respectively. Stop and search involves physical detention and search of a person's body without charge. The evidence shows it falls with severe disproportionality on Black individuals (3.7x the rate of white people, persistent for 20 years). Curtailing its routine use therefore withdraws a coercive power that lands heavily and repeatedly on a specific population. That is a clear O10 improvement. Facial recognition technology used in public spaces constitutes mass biometric surveillance. The Court of Appeal has already found it breached privacy and data protection law. Research links it to documented racial and gender bias, with documented wrongful identifications. Ending routine use would directly reduce the reach of state surveillance into public space, reducing the chilling effect on freedom of movement and assembly — core O10 indicators. The third element — diversity fitness-to-practice assessments for officers — is an internal workforce regulation. It does not restrict citizens' liberty, though it does impose obligations on officers. This is a marginal O10 consideration and does not alter the overall direction. The magnitude is moderate rather than major because: (a) the word 'routine' leaves discretionary or targeted use intact, meaning the liberty gain is partial; (b) the stop and search commitment is to end routine rather than all use; and (c) implementation is uncertain. Confidence is moderate because the O10 mechanism is clear and well-evidenced (surveillance and coercion being withdrawn), but the actual operational scope of 'routine' is undefined in the policy text, which is the central unknown.
Crime, justice & national security — Mixed picture
minor · moderate confidence
Ending routine stop and search and facial recognition may reduce some arrests but evidence suggests both tools have weak crime-reduction effects overall; improved police-community trust could partly offset any operational loss. The net safety effect is small and genuinely contested.
The evidence
- The policy proposes to end the routine use of stop and search and facial recognition software. — greenparty.org.uk (manifesto) — “End the routine use of stop and search and facial recognition software”
- Stop and search has little evidenced deterrent effect on offending. — commonslibrary.parliament.uk (government) — “there is "little evidence to suggest that stop and search provides an effective deterrent to offending"”
- A large majority of stops result in no action — 76% in 2019-20. — theguardian.com (media) — “In 2019-20, 76% of stops led to no further action”
- Only about 20% of searches resulted in a criminal justice outcome linked to the purpose of the search in 2020/21. — researchbriefings.files.parliament.uk (government) — “only about 20% of searches resulted in a criminal justice outcome linked to the purpose of the search”
- Some within policing argue that targeted stop and search can still confiscate dangerous items. — commonslibrary.parliament.uk (government) — “some within policing argue that targeted stop and search can still confiscate dangerous items”
- The Metropolitan Police made approximately 1,300 arrests using live facial recognition between January 2024 and September 2025. — theguardian.com (media) — “1,300 arrests in the Met using LFR from January 2024 to September 2025”
- Ending routine stop and search is predicted to rebuild trust with communities, including by reducing disproportionality that has damaged police-community relations. — libertyhumanrights.org.uk (media) — “Reduced disproportionality is expected to rebuild trust with communities, particularly Black communities, which current practices have "further marginalised" and whose relationships with the police have been damaged”
- The policy also includes ongoing fitness-to-practice diversity assessments for all officers. — greenparty.org.uk (manifesto) — “implement ongoing fitness-to-practice assessments on diversity for all officers”
Biggest unknown: Whether the arrest productivity of facial recognition and targeted stop and search — if properly reformed rather than ended — would exceed the community-intelligence gains from restored trust.
Our reading: For O5 — the protective good only — this policy produces two competing effects. On stop and search: the evidence is clear that the tool has a poor hit rate (76% of stops led to no further action; only 20% linked to a criminal justice outcome) and 'little evidence' of deterrence. Removing its routine use therefore poses a limited direct safety cost. The counterpoint — that targeted use can remove dangerous items — is acknowledged but does not rest on population-scale crime-reduction evidence in the provided units. On facial recognition: the 1,300 Met arrests figure represents a real, if modest, enforcement gain that ending routine FRT use would forego. This is the clearest O5 cost in the evidence. On trust and community cooperation: improved police-community relations — which the policy explicitly targets — can support intelligence flows and voluntary reporting, which are well-established contributors to crime detection. However, the evidence provided for this mechanism is projected rather than measured at scale. On diversity assessments: the evidence on whether such training changes officer behaviour is mixed (some studies show little behavioural effect) and the O5 pathway is indirect. Overall, the policy trades a small but real operational reduction (FRT arrests) against a weak-evidence tool (routine stop and search) and a plausible but unquantified trust dividend. The net O5 effect is minor and genuinely mixed — modest losses in one enforcement channel, modest potential gains via trust — rather than a clear improvement or deterioration.
Equal treatment & democratic rights — Helps
moderate · moderate confidence
This policy directly targets documented racial disproportionality in policing — stopping practices shown to treat Black people unequally — and mandates diversity standards for officers. The main caveat is that the effectiveness of diversity training is disputed and ending stop and search entirely could have trade-offs.
The evidence
- The policy commits to acknowledging institutional racism, misogyny, homophobia, and disablism in policing, ending routine stop and search and facial recognition software, and implementing ongoing fitness-to-practice diversity assessments for all officers. — greenparty.org.uk (manifesto) — “Restore trust and confidence in the police by acknowledging institutional racism, misogyny, homophobia, and disablism. End the routine use of stop and search and facial recognition software, and implement ongoing fitness…”
- Black people are searched at a rate 3.7 times higher than white people across England and Wales. — commonslibrary.parliament.uk (government) — “Black or Black British people were searched at a rate 3.7 times higher than those from a White ethnic group across England and Wales”
- There were 24.5 stop and searches per 1,000 Black people compared to 5.9 per 1,000 white people in the year ending March 2023. — ethnicity-facts-figures.service.gov.uk (government) — “In the year ending March 2023, there were 24.5 stop and searches for every 1,000 Black people, compared to 5.9 for every 1,000 white people in England and Wales”
- This ethnic disproportionality in stop and search has persisted for at least 20 years. — libertyhumanrights.org.uk (media) — “This disparity has been consistently high for 20 years”
- No police force can satisfactorily explain why ethnic disproportionality in stop and search persists. — commonslibrary.parliament.uk (government) — “no force can satisfactorily explain why" ethnic disproportionality persists”
- FRT systems exhibit documented racial and gender bias, with Black women subject to the highest false positive rate (9.9%) in 2025 UK police tests. — post.parliament.uk (government) — “Black women were subject to the highest percentage (9.9%) of false positive identifications at a 0.8 threshold”
- A 2025 retrospective facial recognition test showed higher false positives for Black and Asian individuals. — post.parliament.uk (government) — “A 2025 test on a retrospective facial recognition (RFR) algorithm showed higher false positives for Black and Asian individuals”
- The Court of Appeal ruled in 2020 that South Wales Police's use of live facial recognition breached equality laws. — emergencyservicestimes.com (media) — “The Court of Appeal ruled in 2020 that South Wales Police's use of live facial recognition (LFR) breached privacy, data protection, and equality laws”
- Ending routine stop and search is predicted to rebuild trust with communities that current practices have marginalised. — libertyhumanrights.org.uk (media) — “Reduced disproportionality is expected to rebuild trust with communities, particularly Black communities, which current practices have "further marginalised" and whose relationships with the police have been damaged”
- Ending routine FRT use would reduce discrimination from a technology with documented racial and gender bias. — emergencyservicestimes.com (media) — “By ceasing a technology with documented racial and gender bias, the policy would directly tackle a source of disproportionate policing and potentially wrongful identifications”
- Diversity training effects on officer behaviour are disputed — some studies find little effect on actual behaviour even where knowledge improves. — bps.org.uk (media) — “diversity trainings as they are currently practiced are unlikely to change police behaviour”
- Some evidence-based training programmes have shown positive results in engaging officers. — research.reading.ac.uk (academic) — “other evidence-based programs have shown positive results in engaging officers and fostering open conversations”
Biggest unknown: Whether fitness-to-practice diversity assessments and reduced stop and search actually reduce measurable racial disproportionality in policing outcomes, or whether disproportionality persists through other enforcement routes.
Our reading: O9 concerns equal treatment, anti-discrimination protections, and minority protections. This policy directly addresses documented, persistent, and legally-challenged racial disproportionality across two policing practices — stop and search and facial recognition — that are active sources of unequal treatment for Black and minority ethnic people in England and Wales. The measurable baseline is stark: Black people are searched at 3.7× the rate of white people, a disparity that has lasted 20 years with no force able to explain it. FRT adds a further discrimination vector: UK police tests in 2025 recorded the highest false positive rates for Black women (9.9%), and the Court of Appeal has already ruled one FRT deployment unlawful on equality grounds. Both practices therefore currently worsen O9 in measurable, evidenced ways. Ending routine use of both mechanisms would remove two active sources of discriminatory treatment. The projected gains — restored community trust, reduced wrongful identification — are plausible and linked directly to the equality violations the evidence documents. Unlike many policing reforms, the mechanism here is straightforward: fewer discriminatory contacts means less unequal treatment. The fitness-to-practice diversity assessments are a weaker element. Evidence on whether diversity training changes officer behaviour is genuinely mixed — the British Psychological Society notes current practice is unlikely to change behaviour, though some structured programmes show positive engagement. This component earns only modest weight. Counterfactual: absent this policy, disproportionality in stop and search has persisted for two decades with no force resolving it, and FRT bias has been documented but not corrected. The policy's main instruments — ending routine use of two biased practices — are direct rather than aspirational and would produce measurable reductions in unequal treatment. Magnitude is moderate rather than major: stop and search affects a significant but bounded share of the population, and training uncertainty limits the whole-force culture change the policy implies. Confidence is moderate given the strong baseline evidence but uncertain training effect.