A Guardian article at the weekend reported on the increasing number of people being ejected from stores after being misidentified by facial recognition systems as past shoplifters [Mu26]. This commercial use of facial regulation has even less oversight than police use, which has also been causing alarm. The people at the centre of the report were eventually offered gift vouchers by the shops concerned, but only after considerable personal embarrassment and lengthy and complex processes to clear their names (or to be precise faces).

According to the article Facewatch, the company providing the facial recognition service, claim a 99.98% accuracy rate. This sounds high. Does this mean that the cases reported are rare, albeit unfortunate, incidents?

Let’s unpack this a little.

According to the UK Office of National Statistics annual report on Crime in England and Wales, there are just over half a million cases of shoplifting a year [ONS26]; the Facewatch web site offers a higher figure of 2 million across the whole UK, maybe attempting to take into account underreporting [FW26]. Let’s use this larger figure.

In the UK there are about 55 million adults, assuming on average of one shop visit per day, that is about 20 billion shopping visits per year. So that means shoplifting accounts for just one visit in 10,000.¹

So, if a facial recognition systems said no-one was a past shoplifter, it would attain 99.99% accuracy!² If on the other hand the accuracy is equal for shoplifters and non-shoplifters (that is false positive and false negative rates are the same), then there would be one misidentified innocent for every correctly identified shoplifter — hardly rare. If we use the ONS shoplifting figures, this rises to three misidentifications for each correct one.

One assumes that Facewatch adjusted the system recognition thresholds to have a lower false positive rate (wrongly accused) than this, instead accepting a greater proportion of missed true shoplifters, but in this case an overall 99.98% figure is unachievable. Most likely the reported figure it is based on training data with, perhaps equal numbers of photos of shoplifters and non-shoplifters (essential to allow effective learning), so the 99.98% accuracy figure refers to this data not the numbers of each encountered in realistic (let alone real) use.

In both this case and others, such as rare disease diagnosis, seemingly high stated accuracy rates may not be as good as they at first seem, and certainly need a lot of context to be meaningful. As is clear this is by no means an abstract mathematical discussion, but one that affects real lives. In the case of the use of facial recognition, the article also reminds us that these kinds of systems often have lower accuracy rates, and in particular higher false positive rates (that is wrongly accused) for black and asian people and for women in particular.

References

[FW26] Facewath (2026). Home page. Accessed 4th May 2026. https://www.facewatch.co.uk

[Mu26] Jessica Murray. Guilty until proven innocent: shoppers falsely identified by facial recognition system struggle to clear their names. The Guardian, 3 May 2026. https://www.theguardian.com/technology/2026/may/03/guilty-until-proven-innocent-shoppers-falsely-identified-by-facial-recognition-struggle-to-clear-their-name

[ONS26] Office of National Statistics (2026). Crime in England and Wales: year ending December 2025. ONS Centre for Crime and Justice, 23 April 2026. https://www.ons.gov.uk/peoplepopulationandcommunity/crimeandjustice/bulletins/ crimeinenglandandwales/yearendingdecember2025

It is really hard to keep track of these huge numbers. I’m expert at it, but I initially made a small slip and was out by a factor of 20.[back]
When I read accuracy figures in academic papers on machine learning, I often do the equivalent calculation for a trivial classifier … as in this case, it is often no worse than the algorithm.[back]

In news stories this morning about pet abduction and sustainable fashion, the most critical parts are buried deep in the article: a chance remark that gives away the bigger story.

During the lockdown there has been a steep rise in the cost of dogs and other pets, and this has led to an increase in pet abductions. The most high profile example was when Lady Gaga’s dog walker was shot during the theft of her bulldogs in Los Angeles, but the BBC reports that there are over 2000 pet thefts in the UK alone last year.

Pet abduction to be made new criminal offence in thefts crackdown – BBC News

In principle pet theft is a crime covered by the UK Theft Act, but the use of this evidently does not reflect the emotional harms of pet abductions, hence the need for the new law. Reading further the article says:

Although offences under the Theft Act 1968 carry a maximum term of seven years, ministers say there is little evidence of that being used, because the severity of the sentence is partly determined by the monetary value of the item taken.

It was this that caught my eye. The most severe penalties under the Theft Act are for the most valuable items. If the second-hand car of a pensioner near the poverty line is stolen, it will attract a less severe sentence than the trophy Porsche from the millionaire’s collection. This sounds like a law made in the 17th century, but is in fact from 1968 and applies today.

The lesson is clear, if you are poor then even the criminal law does nothing for you.

The second story is about Molly-Mae, ex-Love Island contestant and social media influencer, who has just been recruited as creative director of Pretty Little Things with a particular focus on sustainable fashion.

Molly-Mae: “I’m not just an influencer anymore”

Reading further there is a section entitled “Wearing the same dress twice”, that has the following quote from Molly-Mae:

“I even captioned one of my Instagram pictures the other day saying ‘PSA it’s ok to wear the same dress twice’ – it’s a bad habit us girls have got into, like if you put it on Instagram it means you can’t wear it again.”

Although I did know some of the figures for this before, it still shocked me to hear that “wearing the same dress twice” is regarded as a significant message.

Sadly, this does reflect the previous figures I’ve seen suggesting that the median number of times a garment is worn is indeed one, with something like 20% of clothes never worn at all once bought. This all has to be added to around 1/3 of fashion clothing that is shredded or otherwise disposed of without ever being sold, due to end of season, returns, or other reasons.

The fashion industry is estimated to contribute 10% of all global carbon emissions, not to mention plastic micro-fibres, chemical, water and other environmental impacts, as well as being built upon near slave-labour conditions across the world.

Given this, even wearing clothes twice could be a major benefit.

However, just imagine how the statement sounds to someone who lived through the second world war, or even anyone over 50. This is reflected in figures for environmental action by age group: awareness is greatest in the younger age groups, but in nearly all areas life-style action is greatest in the older ones. Perhaps influencers such as Molly-Mae can help turn this round.

So as you read the news, do look beyond the headlines, the most hard-hitting parts are often buried deep.

Image: jimbomack66, CC BY 2.0, via Wikimedia Commons