From bans to bodycams: how police facial recognition is moving onto the street
New Capabilities
Axon's Edmonton body-camera pilot tests facial recognition against a 7,000-person watchlist, as the company reverses its 2019 pledge not to put the technology on bodycams.
Axon's Edmonton body-camera pilot tests facial recognition against a 7,000-person watchlist, as the company reverses its 2019 pledge not to put the technology on bodycams.
In December 2025, the Edmonton Police Service began testing Axon body cameras with third-party facial-recognition software. The watchlist contained 6,341 people flagged for risks like 'violent or assaultive,' 'armed and dangerous,' or 'high-risk offender,' plus 724 with serious warrants—roughly 7,000 faces. Officers didn't get real-time alerts; instead, footage was analyzed to test accuracy and workflows.
Three trends converge. Alberta's province-wide body-camera mandate and Axon's reversal of its 2019 pledge not to put facial recognition on bodycams come together in this pilot—its ethics board had warned of bias and civil-rights risks. Global rules are fracturing: the EU bans real-time biometric surveillance, the UK leans in, North America has bans, lawsuits, and aggressive deployments.
People on Edmonton’s facial-recognition watchlists
6,341 individuals with EPS “flags or cautions” (e.g., violent, armed, escape risk, high‑risk offender) plus 724 with serious outstanding warrants targeted by the pilot.
6+ years
Time since first major U.S. city ban
San Francisco banned municipal police use of facial-recognition technology in May 2019, catalyzing a wave of local restrictions and public debate.
1,400+
Arrests tied to live facial recognition in London
Between early deployments and September 2025, the Metropolitan Police Service in London reports over 1,400 arrests linked to Live Facial Recognition operations, used to justify nationwide expansion.
20+ billion
Images in Clearview AI’s faceprint database
Clearview AI amassed tens of billions of images scraped from public websites, prompting regulator findings of “mass surveillance” and major settlements under Illinois’ biometric privacy law and European data‑protection rules.
EU-wide ban (with narrow exceptions)
Legal status of real-time police facial recognition in the EU
The EU’s AI Act designates real‑time remote biometric identification in public spaces for law‑enforcement as a prohibited practice, allowed only in strictly defined, judge‑authorized situations such as terrorism or serious violent crime.
Voices
Curated perspectives — historical figures and your fellow readers.
Ever wondered what historical figures would say about today's headlines?
Sign up to generate historical perspectives on this story.
15 events
Latest: December 7th, 2025 · 6 months ago
Showing 8 of 15
JK to step
Tap a bar to jump to that date
Jump to
December 2025
Global spotlight on Edmonton pilot as test case for police facial-recognition bodycams
LatestMedia Investigation
Associated Press and partner outlets publish an in‑depth report on the Edmonton pilot, highlighting Axon’s shift from its 2019 stance, concerns from former ethics advisors like Barry Friedman, the lack of transparency about the facial-recognition vendor, and the potential for the trial to pave the way for U.S. deployments.
Edmonton switches on AI-powered watchlist scanning for ~7,000 people
Technology Deployment
The facial-recognition bodycam pilot goes live, with cameras trained to detect faces matching a roughly 7,000‑person high‑risk watchlist derived from EPS flags and serious warrants. Matches are processed after the fact; officers do not receive live alerts during this phase.
Edmonton Police announce facial-recognition bodycam proof of concept
Technology Deployment
EPS issues a news release declaring it will test facial‑recognition‑enabled body‑worn video cameras in partnership with Axon. Up to 50 officers already using bodycams will receive upgraded units for December to assess feasibility and functionality.
October 2025
London’s Met Police reports 1,400+ arrests from Live Facial Recognition
Technology Impact Report
The Metropolitan Police publishes an annual report stating that from September 2024 to September 2025, LFR deployments produced 962 arrests, bringing the cumulative total above 1,400, including many for serious violence and sexual offenses, bolstering government arguments for national expansion.
September 2024
Edmonton rolls out bodycams service-wide
Technology Deployment
EPS begins equipping about 280 officers with Axon bodycams as part of a phased, service‑wide rollout across multiple divisions and specialized teams, making cameras a routine part of front‑line policing.
June 2024
EU AI Act agreement cements strict limits on police biometric surveillance
Legislation
EU institutions reach political agreement on the AI Act, including a near‑total ban on real‑time remote biometric identification in public spaces for law enforcement, with narrow, judge‑authorized exceptions for specified serious crimes and threats.
February 2024
RCMP starts Alberta bodycam field tests
Technology Deployment
The Royal Canadian Mounted Police begins field‑testing body‑worn cameras and a digital evidence management system in several Alberta detachments, using the results to finalize a national rollout that will likely rely heavily on Axon hardware and cloud tools.
July 2023
Edmonton Police Service begins body-worn camera proof of concept
Technology Deployment
EPS launches a proof of concept with select officers to evaluate operational impacts of body‑worn video, ahead of a province‑driven service‑wide rollout.
March 2023
Alberta mandates body-worn cameras for provincial police agencies
Policy
The Government of Alberta announces that all municipal and First Nations police services, along with Alberta Sheriffs, must adopt body‑worn cameras, citing goals of transparency, complaint resolution, and evidence quality.
June 2022
Axon ethics board resigns over Taser-armed drone plan
Corporate Governance Crisis
Nine of 13 members of Axon’s AI Ethics Board resign after the company announces plans to deploy Taser‑equipped drones and surround schools with surveillance cameras, warning of mission creep and racialized harms. Axon pauses the drone project but later replaces the board with a less transparent advisory council.
May 2022
Clearview AI settles ACLU lawsuit, limiting private-sector sales of face database
Settlement
Clearview AI agrees to a landmark settlement under Illinois’ Biometric Information Privacy Act that permanently bans it from providing its faceprint database to most private entities in the U.S., though law‑enforcement customers remain exempt.
February 2021
Canadian regulators deem Clearview AI’s practices unlawful mass surveillance
Regulatory Action
The Office of the Privacy Commissioner of Canada and provincial counterparts rule that Clearview AI’s scraping of billions of facial images without consent violates federal and provincial privacy laws, labeling it mass surveillance and ordering it to stop operating in Canada.
August 2020
UK court rules early live facial-recognition deployments unlawful
Judicial Ruling
In Bridges v. South Wales Police, the Court of Appeal finds the force’s LFR deployments breached privacy rights and equality‑duty obligations due to broad watchlists and failure to assess algorithmic bias, sending a warning shot to UK police using similar tools.
June 2019
Axon ethics board advises against facial recognition on bodycams; company agrees
Corporate Decision
After a year of study, Axon’s AI & Policing Technology Ethics Board concludes facial recognition is not reliable or fair enough for body‑worn cameras and warns of unequal performance across races and genders. Axon publicly commits to keep face recognition off its products.
May 2019
San Francisco becomes first major U.S. city to ban police use of facial recognition
Legislation
San Francisco’s Board of Supervisors votes 8–1 to ban city agencies, including police, from using facial-recognition technology, citing civil‑rights risks and accuracy concerns. The Stop Secret Surveillance Ordinance becomes a model for other municipal bans.
Historical Context
3 moments from history that rhyme with this story — and how they unfolded.
1 of 3
2019–2021
Municipal bans on police facial recognition in U.S. cities
Starting with San Francisco in May 2019, several U.S. cities—including Somerville, Oakland, and Boston—enacted ordinances banning city agencies from using facial-recognition technology, often as part of broader surveillance-oversight laws. Advocates cited racial bias, risk of wrongful arrests, and the danger of pervasive government tracking in public spaces.
Then
Local police departments in these jurisdictions halted or avoided deploying facial recognition, and the bans sparked national debate over the technology’s role in law enforcement.
Now
The municipal bans became reference points for civil‑liberties campaigns and helped legitimize arguments for stronger state and federal biometric privacy laws, even as other jurisdictions continued or expanded facial-recognition use.
Why this matters now
Edmonton’s bodycam pilot represents a counter‑trend to city‑level bans: instead of withdrawing from facial recognition, a major North American police service is experimenting with putting it directly onto officers’ bodies. The contrast underscores how, absent national rules, local political cultures and vendor relationships heavily shape whether communities experience biometric policing as normalized infrastructure or prohibited practice.
2 of 3
2017–2020
Bridges v. South Wales Police – early live facial recognition ruled unlawful
South Wales Police deployed live facial-recognition cameras at events and busy streets, scanning crowds against watchlists of suspects and persons of interest. Civil-rights campaigner Ed Bridges challenged the practice. In August 2020, the UK Court of Appeal held that deployments violated privacy rights under the European Convention on Human Rights and breached the public-sector equality duty because police left officers too much discretion and failed to assess algorithmic bias.
Then
South Wales Police adjusted its policies, and the ruling cast doubt on similar deployments by other UK forces, but did not end live facial recognition; usage resumed under revised guidance.
Now
Bridges became a touchstone case showing that courts can rein in biometric surveillance on proportionality and discrimination grounds, even without explicit facial-recognition statutes—an approach that could be mirrored in future challenges to bodycam‑based facial recognition.
Why this matters now
The case illustrates how legal standards around necessity, proportionality, and bias can determine whether a particular form of facial recognition is acceptable. Edmonton’s pilot—scanning the public with mobile cameras—raises similar questions about scope of watchlists, discretion, and fairness that courts or regulators may eventually apply in Canada or elsewhere.
3 of 3
2019–2025
Clearview AI investigations and settlements over mass face scraping
Clearview AI scraped billions of images from social-network and web platforms to build a massive facial-recognition database, which it sold to police and private clients. Lawsuits and regulatory probes in the U.S., Canada, and Europe alleged violations of biometric and data‑protection laws. Canadian regulators labeled its practices unlawful mass surveillance and forced it out of the market; in Illinois, an ACLU‑led lawsuit produced a settlement barring most private‑sector access to its face database; in the EU, data‑protection authorities have issued large fines and deletion orders.
Then
Clearview became a high‑profile cautionary tale, prompting some agencies to pause or formalize their facial-recognition use and demonstrating that strong biometric laws like Illinois’ BIPA have real teeth.
Now
The company’s partial survival—continuing to serve law‑enforcement clients despite restrictions—shows that litigation tends to shape how facial recognition is provided rather than banning it outright. It also underscores the power of robust legal frameworks and privacy commissioners in constraining abusive models.
Why this matters now
Edmonton’s bodycam trial differs technically from Clearview—it uses an internal mugshot database rather than scraped social‑media images—but it raises similar structural concerns: dragnet scanning of people in public, unclear error rates, and limited avenues for those on watchlists to know and challenge their inclusion. The Clearview saga suggests that, if regulators and courts view bodycam facial recognition as disproportionate or opaque, they may impose strong conditions or prohibitions even without waiting for new, bespoke legislation.
