Facewatch Live: Free Retail Face Recognition Tool [2026]

Update — November 4, 2026: The Home Office announced a police facial recognition law and oversight body; retail use stays under data protection law for now. See UK police facial recognition law and oversight body.

On October 27, 2026, Biometric Update reported that Facewatch has begun live beta testing of Facewatch Live, a free retail crime intelligence platform. Shops can use it to document incidents, flag repeat offenders and build structured case files for police, either on its own or alongside Facewatch's live facial recognition system. The article says Frasers Group is among the first retailers testing Facewatch's system in daily operations, without specifying which product. The news matters because "free" lowers the barrier: a platform that costs nothing to adopt can spread to far more shops, and each one that pairs it with facial recognition adds shoppers' faces and unproven allegations to someone's database.

If you shop in the UK, appear in other people's store or street footage, or publish video that catches bystanders, this affects you. This guide covers what Facewatch Live does, what happened when a Sainsbury's shopper was wrongly flagged, what your rights are, and what you can do today, including how to blur faces in footage you share.

What Is Facewatch Live and What Is New About It?

Facewatch Live is a free tool that lets a retailer log theft and abuse incidents, link several incidents to the same offender profile, and package them into case files for police. According to Biometric Update's report, it "can operate independently of the company's live facial recognition system or alongside it", so a retailer can choose incident reporting alone or combine it with biometric identification.

Facewatch chief executive Nick Fisher is quoted as saying: "Retail operators should not have to pay simply to understand what is going on in their own businesses." That is the commercial logic: give away the incident log, and sell the facial recognition on top.

What is genuinely new is the packaging, not the idea. Retailers have always kept incident books. What changes is that:

• Offender profiles are linked across incidents, so a person can become a "repeat offender" record.
• Evidence is compiled into case files structured for police submission.
• Biometric matching can be layered on when the shop also subscribes to live facial recognition.

The article notes that even without facial recognition, these records could hold CCTV images, suspected offender profiles and theft allegations. It also notes that the ICO states that data protection rights extend to suspected offences and unproven allegations. Biometric Update gives no timeline for the beta and no ICO enforcement action.

How Does Facewatch Facial Recognition Work in a Shop?

In a live facial recognition deployment, cameras at the entrance capture faces, software converts each face to a numerical template, and the template is compared with a watchlist of people a retailer has reported. If it matches, staff receive an alert, usually on a phone.

The Register's reporting on Sainsbury's gives the best public picture of how the claims and the reality line up. The Register reported that Facewatch reports 99.98% accuracy and a 46% drop in logged incidents of theft, harm, aggression and antisocial behaviour, and says 92% of offenders do not return to stores using the system. Grocery Gazette adds that Sainsbury's said in July 2026 that people review all alerts before action is taken and that images of unmatched shoppers are deleted instantly.

Those are the companies' own figures. They have not been independently audited in anything we found, and a high accuracy percentage can still mean many wrong flags when millions of faces pass through doors. Treat them as claims, not as measured outcomes.

Why does a "match" still need a human?

Because the software produces a similarity score, not a fact. The human reviewer is the safeguard the companies point to. The Sainsbury's case below shows where that safeguard can fail: after the alert, in the shop, between a phone screen and a real person.

What Happened When Sainsbury's Wrongly Approached a Shopper?

Warren Rajah, who works in sales, was approached by a manager at the Sainsbury's in Elephant and Castle, London, and told to leave, The Register reported on February 6, 2026. The facial recognition system had alerted staff to a match on the offenders' database. Sainsbury's said it had correctly identified a man on that database, but that the manager approached the wrong person. Three managers reportedly looked at their phones, then at Rajah, before asking him to leave.

Facewatch called it "a case of human error in-store" and said its data protection team confirmed Rajah had no alerts. Sainsbury's apologised, said staff would receive further training, and said the problem was not the technology.

The part that matters for rights is the aftermath. To clear his name, Rajah had to submit a copy of his passport and a headshot to Facewatch, which then confirmed he was not on the database. Big Brother Watch told The Register that innocent people are often offered little more than an apology and must hand over more personal data to clear themselves. The group is challenging live facial recognition in the courts.

Whether you call it a technology failure or a process failure, the shopper bore the cost: public humiliation first, and more personal data second.

How big is the rollout?

The Register said the system started in September 2025 in Sydenham and Bath Oldfield Park and had reached six Sainsbury's stores by February. Grocery Gazette reported in July that it was live in 55 stores, with plans for up to 150 more before Christmas. That growth is why a free add-on from the supplier is notable: the market is expanding fast. The Register lists other Facewatch users as B&M, Budgens, Costcutter, Southern Co-op, Spar and Sports Direct, and mentions Iceland as having trialled it.

Is Retail Facial Recognition Legal in the UK?

It is currently used, but its legality depends on how each deployment meets UK data protection law. Facial templates used to identify a person are biometric data, a special category under the UK GDPR, which means a retailer needs a lawful basis, an additional condition for special category data, a data protection impact assessment, and clear notice. Our guide to UK GDPR and the Data Protection Act for video privacy walks through those duties for anyone filming people.

Three points make this contested:

1. Proportionality. Retailers say shoplifting justifies the tool. Campaigners say scanning every entrant is disproportionate when most are innocent.
2. Accuracy and bias. Misidentification falls on real people, and error rates can differ across groups.
3. Redress. There is no simple route for a wrongly flagged shopper to correct a record or learn who is on a watchlist.

The wider UK picture is moving. Police live facial recognition is being extended, as in permanent cameras on Oxford Street, and the Home Office has published a DPIA covering live facial recognition for immigration enforcement. Regulators have also been tested in court over face databases, as in the Clearview AI UK Court of Appeal ruling. Private retail use sits alongside all of this with less public debate.

What Are the Risks of a Free Incident Platform Without Facial Recognition?

Even with no face matching, an incident log can hold CCTV stills and a named or described suspect, which is personal data about an accusation. Three risks stand out:

• Unproven allegations become records. A person flagged in one store's notes may be linked to "the same offender" in another incident on weak evidence.
• Sharing. Case files designed for police submission raise the question of who else sees them and for how long they are kept.
• Scope creep. A free tool that starts as an incident book can become the on-ramp to a paid biometric subscription.

The source article does not say how long Facewatch Live keeps data or who can see it. Until that is published, treat those as open questions rather than assumed harms.

What Are Competing Retail Platforms Doing?

Biometric Update notes that Asda plans to deploy Auror Core across its stores after a 30-store trial in London, and that the College of Policing documented Auror's direct police reporting in a Devon and Cornwall Police deployment, where crimes reported through the platform had higher charge rates and shorter reporting times. The pattern is clear: retail crime intelligence is becoming a category, and police pipelines are a selling point.

What People Are Saying

We did not find a Hacker News thread tied to this specific launch, so there is no community discussion to quote. The reaction we can cite is from the reporting above: Big Brother Watch's criticism of the burden on wrongly flagged shoppers, and Rajah's description of the experience as mentally debilitating for vulnerable people.

What Should You Do If You Are Wrongly Flagged or Worried?

Act on records and rights, and keep evidence. Here is a practical order:

1. Look for signage. Shops using live facial recognition should say so at the entrance and in a privacy notice. Photograph the sign if you have a concern.
2. If you are challenged, stay calm and ask why. Ask what system flagged you, and ask for the company name and the store manager's name. Note the date, time and store.
3. Send a subject access request. Write to the retailer and to the provider asking for all personal data held about you, including images and incident records. Normally you should expect a reply within one month.
4. Ask what ID is truly needed. Rajah had to supply a passport and headshot. Ask whether a lower-risk proof will do, and who will hold the copy.
5. Complain to the retailer, then to the regulator. Put the complaint in writing and, if unresolved, raise it with the Information Commissioner's Office.
6. Consider support groups. Organisations such as Big Brother Watch collect cases of misidentification.
7. Keep your own footage clean. If you film in public places and shops, blur bystanders before you publish.

How Can Creators Blur Faces in Store and Street Footage?

If you post video that catches other shoppers or passers-by, blurring their faces reduces the chance that they are identified by viewers or by face search tools. Retail and public footage is also what many people share when they are documenting an incident or a store's practices.

Here is how to do it with BGBlur, which runs in your browser:

1. Upload your clip (MP4, MOV or M4V).
2. Choose face blur. The AI detects faces and tracks them as people move, so you do not edit frame by frame.
3. Check the whole clip. Scrub through crowded sections and edges of the frame where small or turned faces can be missed, and fix any gaps.
4. Add license plate blur if vehicles outside are visible.
5. Export in MP4, MOV or WebM.

On pricing, the Free plan covers 200MB, 10 minutes and 3 videos a month at 720p export; paid plans add more. For more on the principles, see our guide to blurring faces in photos and videos with AI and our explainer on unauthorized filming and face blurring. If you capture footage from a doorbell or home camera, the bystander footage guide applies as well.

What blurring cannot do

Be honest about the limit. Blurring a video you publish does nothing about a retailer's own cameras, the watchlist, or a facial template created at the door. It protects the people in your footage from being identified through your upload. It is also not a reason to skip consent where the law requires it. And weak or light blur can sometimes be reversed, so use a strong setting; our piece on forensic de-pixelation explains why.

What Should Retailers and Businesses Consider?

If you run a shop, the same law applies to you whether the tool is free or paid. Before enabling any watchlist or case-file feature:

• Complete a data protection impact assessment and keep it current.
• Post clear, visible signage and a plain-language notice.
• Define retention periods and who may access incident records.
• Build a fast correction route for people who are wrongly flagged, without demanding passports as the default.
• Train staff so that an alert is verified against the actual person before anyone is approached. The Sainsbury's case turned on that step.
• If you publish your own CCTV clips for appeals or marketing, blur the faces of uninvolved customers first. Our CCTV processing tools comparison is a starting point.

Update — October 28, 2026: In the US, a report on Kroger's Giant Eagle deal raises the same retail camera questions; see Kroger–Giant Eagle and Flock cameras.

What Happens Next?

Three things are worth watching. First, the Facewatch Live beta: its data retention terms, sharing rules and how many retailers adopt it. Second, Big Brother Watch's legal challenge to live facial recognition. Third, whether the ICO or Parliament sets clearer rules for private-sector watchlists. None of these has a published date in the sources we reviewed, so this post will be updated if they move.

For readers outside the UK, the principle travels. In the EU, retrospective facial recognition on CCTV archives is subject to AI Act scrutiny, and the same blurring principles apply to any footage you publish.

Conclusion: Your Face, Their Database

Facewatch Live is a small product announcement with a large implication. By making incident logging free and pairing it with live facial recognition, Facewatch is lowering the cost of building watchlists of suspected offenders, in a year when Sainsbury's alone has gone from a handful of stores to 55 with more planned. The company's headline figures, 99.98% accuracy and a 46% fall in incidents, are its own, and the Rajah case shows what a single error costs the person on the receiving end: being ejected in public, then handing over a passport to prove innocence.

You cannot control a shop's cameras, but you can control your rights and your own footage. Know the signage, use subject access requests, keep records if you are challenged, and complain in writing. When you film and share, blur the people who did not agree to be in your video. Try BGBlur's face blur on your next clip: it runs in your browser, tracks faces automatically, and lets you check the result before you publish.