AI Cameras Detect Driver Violations in the UK: What They Catch and What the Law Says
Key Takeaways
- The BBC reported on 15 November 2026 that a two-week AI camera trial in the West Midlands uncovered nearly 19,000 potential driving offences, including a driver drinking from a wine bottle and another inhaling from balloons.
- The same technology is being trialled in Devon and Cornwall to detect drivers who may be impaired by drink or drugs, in what police describe as a world-first trial.
- The cameras do not issue fines or convict anyone. They flag images for human review, or alert officers down the road, who then stop the vehicle and carry out roadside tests.
- The offences they catch carry serious penalties: using a handheld phone means a 200 pound fine and six penalty points, not wearing a seatbelt can mean a fine of up to 500 pounds, and drink driving carries a minimum 12-month ban.
- The West Midlands trial involved no enforcement at all. It was a test to measure the scale of the problem, funded by the West Midlands Combined Authority with police and fire service involvement.
- Privacy questions remain open: how long images are kept, who can see them, and what happens when the system flags an innocent driver.
A driver drinking from a bottle of wine at the wheel. A front-seat passenger holding a small unrestrained child. Another driver inhaling from balloons. These are not scenes from a road safety campaign film. They are images captured by AI cameras during a two-week trial in the West Midlands, reported by the BBC on 15 November 2026, and they have put the technology back at the centre of a national argument about how Britain polices its roads.
The trial found nearly 19,000 potential driving offences in fourteen days at just two locations. This article explains what the cameras can now detect, what the law says about the offences they catch, and what happens after a camera flags a driver, focusing on enforcement and drink-driving rather than the general technology, which we have already explained in AI Video Surveillance Explained.
The Image That Started the Conversation
The BBC report, published on 15 November 2026, described an artificial intelligence trial that uncovered nearly 19,000 potential driving offences in two weeks. AI-powered cameras installed outside Ladywood fire station in Birmingham and Fallings Park in Wolverhampton captured drivers holding mobile phones at the wheel, failing to wear seatbelts, and, in the most striking cases, drinking from a bottle of wine and inhaling from balloons while driving.
Mat MacDonald, the West Midlands Road Safety Commissioner, said he was not surprised by what the cameras found in the March trial, and singled out one photograph of a driver drinking from a bottle of wine as particularly galling. He told the BBC he is working to make the technology a reality on the region's roads and is hopeful of a wider rollout.
The trial was actually stopped earlier than planned because of how many possible road offences were being detected. That detail matters, because it suggests the scale of the problem surprised even the people running the test. Road safety data from 2025, cited by the West Midlands Combined Authority, showed mobile phone use was linked to 32 deaths on UK roads, while unfastened seatbelts were linked to 133 deaths.
Simon Foster, the West Midlands police and crime commissioner, previously said the findings were being closely evaluated to consider how the technology could be deployed for future police enforcement. The message from both men is the same: the trial was a measurement exercise, and the measurement was alarming enough to justify talking about enforcement.
The Numbers: Nearly 19,000 Offences at Two Sites
The breakdown of the West Midlands figures shows where the problem lies.
At Ladywood, the cameras identified 11,045 potential seatbelt offences and 1,163 potential mobile phone offences. At Fallings Park, they identified 6,030 potential seatbelt offences and 757 potential mobile phone offences. Added together, that is 17,075 potential seatbelt offences and 1,920 potential mobile phone offences, a total of 18,995 potential offences in two weeks at two locations.
Seatbelt offences dominate the count by a wide margin, outnumbering phone offences by roughly nine to one. That is itself a significant finding. Public debate about dangerous driving tends to focus on phones, which are visible and widely condemned. The cameras show that failing to wear a seatbelt is far more common, and it is the offence linked to far more deaths in the 2025 data: 133 deaths linked to unfastened seatbelts against 32 linked to mobile phone use.
The word "potential" does a lot of work in these figures and should not be skipped over. The cameras flag images that appear to show an offence. A human reviewer must then confirm whether an offence actually occurred. A seatbelt hidden by clothing, a phone held by a passenger, or an image taken at an angle that obscures the driver's hands can all produce a false flag. The trial numbers are therefore a measure of what the cameras detected, not a count of proven offences, and they are not the same as 19,000 fines or prosecutions.
No enforcement was planned as a result of the trial, which was funded by the West Midlands Combined Authority working with the police and fire service. Nobody captured on camera in Birmingham or Wolverhampton received a fine from this trial. That is an important fact for anyone who worries about being prosecuted by a machine: in this case, the machine was only watching and counting.
Our earlier report on the same trial, AI Cameras Caught 19,000 UK Drivers in Two Weeks: How They Work, covers the technology in more detail. This article focuses on what comes next: the drink-driving detection, the law, and the privacy questions.
The New Frontier: Detecting Drink and Drug Drivers
The West Midlands trial caught drivers using phones and skipping seatbelts. A separate trial in the south-west of England is trying to do something harder: spot drivers who may be impaired by drink or drugs before they cause a crash.
Devon and Cornwall Police are trialling AI cameras developed by the Australian firm Acusensus that analyse images of passing cars for signs of driver impairment. The technology, called Heads Up, is an extension of the same system used to detect phone use and seatbelt offences. The trial is part of the Vision Zero South West road safety campaign, which aims to reduce the number of people killed or seriously injured on the region's roads.
The system works differently from a breathalyser. It does not measure alcohol in a driver's body. It analyses image data from passing cars, looking at vehicle and driver movements over a period of time, and flags behaviour consistent with impairment, such as swerving or erratic braking. If the system detects possible impairment, it alerts officers stationed down the road, who then stop the car and perform roadside testing to check whether the driver is under the legal limit.
Geoff Collins, Acusensus's UK general manager, has been careful to explain the limits of the system. The AI is not making final decisions about impairment. It is designed to mimic the instincts of experienced traffic officers, identifying the same patterns a trained officer would notice. As Devon and Cornwall Police's head of road safety, Adrian Leisk, put it on the BBC's Crimewatch Live programme, it is down to the officer who stops the vehicle to determine whether the driver is impaired and to conduct the procedures that follow.
The trial was first run over the Christmas period in December 2024, deliberately timed to coincide with the busy Christmas party season, when drink-driving risk is at its highest. During two live trials in December, four impaired driving offences were detected, leading to four drivers being charged. Acusensus has described the early results as very encouraging, although specific data on subsequent tests has not yet been made public by the police.
The same technology has been used in other parts of the country. In Greater Manchester, the cameras were trialled over a five-week period and caught 3,200 people using their phones or not wearing seatbelts, and nine police forces across the UK have trialled the system.
The drink-driving application is described as a world-first trial, and it matters because it changes the role of the camera. A phone or seatbelt camera documents an offence that is visible in a single image. An impairment camera makes a judgement about a driver's condition from patterns of behaviour, and that judgement is then tested by a human officer with a breathalyser or drug test. The camera does not convict. It directs human attention to the drivers most worth stopping.
What the Law Says: Phones
Using a handheld mobile phone while driving is a criminal offence in the UK, and the law is stricter than many drivers realise.
It is illegal to hold and use a phone, sat nav, tablet, or any similar device capable of sending or receiving data while driving. The restriction applies even when the vehicle is stopped at traffic lights or in a queue, because the law treats the driver as driving whenever the engine is running. The only time a driver can legally use a handheld device is when safely parked with the engine switched off.
Since March 2022, the offence covers almost any use where the driver is holding the device, including offline use such as scrolling through a playlist or checking a map. Before that change, prosecutors had to prove interactive communication, such as a call or a message. That requirement has been removed, which means holding a phone in flight mode is still an offence.
The standard penalty is a 200 pound fixed penalty notice and six penalty points on the licence. If the matter goes to court, the fine can reach 1,000 pounds, or 2,500 pounds for a driver of a bus or lorry, and the court can impose a driving disqualification in serious cases. For a new driver who passed their test within the last two years, six points means an automatic loss of the licence under the new drivers rules.
Hands-free use is not prohibited in the same way, provided the device is secured in a cradle and the driver does not hold it. That does not mean hands-free use is risk-free. Research has found that reaction times when using a hands-free device can be around 30 percent slower than reaction times at the drink-drive limit, and the House of Commons Transport Select Committee has said hands-free devices carry a similar risk of an accident to using a phone at the wheel. The law draws a line at holding the device. Safety evidence suggests the risk does not stop at that line.
Accumulating 12 or more penalty points within three years normally leads to a minimum six-month driving disqualification under the totting-up rules. Two phone offences, at six points each, are enough to reach that threshold.
What the Law Says: Seatbelts
Seatbelt law is simpler than phone law, but the penalty is often misunderstood. Drivers and passengers must wear seatbelts where they are fitted, with limited exemptions for medical reasons, delivery drivers reversing, and certain other narrow cases. Children must use the correct child seat or restraint for their size and weight.
A driver who does not wear a seatbelt when required can be fined up to 500 pounds. The driver is also responsible for ensuring that children under 14 are properly restrained, and can be fined for failing to do so. Adult passengers are responsible for their own seatbelts.
The West Midlands trial's most disturbing image, beyond the wine bottle, was a front-seat passenger holding a small child who was unrestrained with no seatbelt. That is not only dangerous. It is the kind of image that explains why road safety campaigners regard seatbelt enforcement as a priority, even though it attracts less public attention than phone use.
Many drivers treat seatbelt offences as minor, because the fine is lower than the phone penalty and no points are added. The casualty figures suggest the opposite: unfastened seatbelts were linked to 133 deaths in the 2025 data cited by the West Midlands Combined Authority, more than four times the number linked to phone use.
What the Law Says: Drink and Drug Driving
Drink driving is the most serious of the offences the cameras are now being used to detect, and the penalties reflect that.
In England, Wales, and Northern Ireland, the legal alcohol limit is 80 milligrams of alcohol per 100 millilitres of blood, or 35 micrograms per 100 millilitres of breath, or 107 milligrams per 100 millilitres of urine. In Scotland, the limit is lower: 50 milligrams per 100 millilitres of blood, or 22 micrograms per 100 millilitres of breath. There is no reliable way to convert these limits into a number of drinks, because the effect of alcohol depends on body weight, sex, metabolism, and food intake. The only certain way to stay legal is not to drink before driving.
Driving or attempting to drive while over the limit carries up to six months in prison, an unlimited fine, and a driving ban of at least 12 months. The ban rises to three years if the driver is convicted twice in ten years. Being in charge of a vehicle while over the limit, which can mean sitting in the driver's seat with access to the keys even without driving, carries up to three months in prison, a fine of up to 2,500 pounds, and a possible ban. A drink-driving conviction is recorded with the DR10 endorsement code, which stays on the licence for 11 years and typically pushes insurance costs up sharply.
Drug driving carries similar penalties: a minimum 12-month ban, an unlimited fine, and up to six months in prison. The law sets specified limits for a list of controlled drugs, and it is an offence to drive above those limits whether or not the driver's ability is visibly impaired.
The government published a Road Safety Strategy in January 2026 that proposed bringing the limit in England and Wales down to the Scottish level, 50 milligrams per 100 millilitres of blood. That proposal has not become law, and the higher limit remains in force in England and Wales as of November 2026. Drivers should not assume the law has changed until Parliament changes it.
Two points about the AI drink-driving trials follow from this legal framework. First, the camera cannot measure blood alcohol. It can only flag behaviour that suggests impairment. The legal test is the breath, blood, or urine specimen taken by police, not the camera's assessment. Second, a driver flagged by a camera and then found to be under the limit has committed no offence, and the stop itself is lawful because police already have the power to stop any vehicle and to require roadside tests where they suspect impairment. The camera changes which cars get stopped. It does not change what counts as an offence.
The Privacy Questions Nobody Has Fully Answered
The technology raises privacy questions that are legitimate and have not yet been answered in full published detail.
The cameras photograph the inside of private vehicles, including passengers who may have committed no offence. A front-seat passenger holding a child, a driver eating lunch, or a passenger in a private conversation can all be captured in images that were taken to detect driving offences. The question of how long those images are kept, who can access them, and how they are deleted is central to whether the system is proportionate.
The legal framework exists in outline. The use of surveillance cameras by public authorities in the UK is governed by data protection law, including the UK General Data Protection Regulation and the Data Protection Act 2018, and by the Surveillance Camera Code of Practice. Police forces must have a lawful basis for processing the images, must limit collection to what is necessary, and must keep personal data no longer than needed. But the general framework does not answer the specific questions: how many days are images retained, are images of innocent drivers deleted automatically, and can the data be used for purposes other than road safety?
Accuracy is a second concern. The system will inevitably produce false positives, flagging drivers who were not committing an offence. In a trial, a false positive is a statistic. In an enforcement deployment, a false positive is a letter through the door, a potential fine, and a demand on the driver's time to contest it. The safeguard is human review, but the quality of that safeguard depends on how carefully reviewers check flagged images and how easy it is for drivers to challenge a decision.
A fourth concern is fairness. If the system's training data is not representative of all drivers, vehicles, and conditions, it may flag some groups more often than others for the same behaviour. No published data we could find breaks down the trial results by driver demographics, so this question cannot be answered from the available evidence. It should be asked before any national rollout, not after it.
These are not arguments against the technology. The casualty figures, 32 deaths linked to phones and 133 to seatbelts in the 2025 data, are a strong argument in its favour. They are arguments for building the governance before the rollout, so that the public can trust a system that photographs them every time they drive past a camera.
Where This Goes Next
Several developments will determine how AI cameras reshape UK driving enforcement.
The first is the West Midlands decision. Mat MacDonald has said he is working to make the technology a reality on the region's roads, and Simon Foster has said the trial results are being evaluated for future police enforcement. If the West Midlands moves from trial to enforcement, it will become the first test of whether the public accepts camera-based enforcement of seatbelt and phone offences at scale.
The second is the Devon and Cornwall drink-driving trial. If the impairment detection proves reliable, and if the number of impaired drivers caught justifies the cost, other forces are likely to adopt it. The trial's success will be measured not by how many drivers are flagged, but by how many are found to be over the limit after being stopped, and by whether the system catches drivers that officers on patrol would have missed.
Drivers can expect to see more of these cameras, not fewer. The direction of travel is clear from the number of forces trialling the system and from the scale of the West Midlands results. What is not yet clear is the governance framework that will accompany them, and that is the part of the story worth watching most closely.
The Bottom Line
AI cameras in the UK now detect more than phones and seatbelts. They are being tested as a way to spot drivers who may be impaired by drink or drugs, alerting officers who then carry out the tests that the law requires. The West Midlands trial showed the scale of the problem the technology is aimed at: nearly 19,000 potential offences in two weeks at two sites, including a driver drinking wine at the wheel.
The law behind the cameras is severe, and it should be. A phone offence means a 200 pound fine and six points. A seatbelt offence can mean a fine of up to 500 pounds. Drink driving means a minimum 12-month ban, an unlimited fine, and up to six months in prison. The cameras do not change those penalties. They change how likely a driver is to be caught.
The technology is not a judge, and it should not be treated as one. It flags images for human review, or alerts officers who make their own assessment. That human step is the safeguard that makes the system defensible, and it is the part that must be protected as the cameras spread. The question for the UK is not whether AI cameras work. The evidence from Birmingham, Wolverhampton, Devon, and Cornwall suggests they do. The question is whether the country builds the rules, the oversight, and the data protections to match before the cameras become a permanent feature of every major road.
What We Could Not Verify
- We could not verify the exact dates of the West Midlands trial beyond the BBC's statement that it took place in March, because the BBC report does not give the year in the version we read, and we have not assumed a year. The trial findings were reported on 15 November 2026.
- We could not verify how long images captured by the cameras are retained, who can access them, or the process for deleting images of drivers who committed no offence, because no published source we could find sets out those details for the trials described.
- We could not verify the accuracy rate of the impairment detection system, including how often it flags drivers who turn out to be under the limit, because the police have not published that data.
- We could not verify whether the drink-drive limit in England and Wales will be lowered to the Scottish level, because the January 2026 Road Safety Strategy proposal has not become law as of November 2026, and its future depends on Parliament.
- We could not verify the Greater Manchester figure of 3,200 offences beyond the secondary reporting cited, and we have attributed it accordingly.
Sources
- BBC News, via cloudfront: Driver drinking from wine bottle among shocking images captured by AI cameras, https://d33fn5lypxzr2z.cloudfront.net/news/articles/cm4gndpl1nlxo
- NewsBytes: West Midlands AI camera trial logged nearly 19,000 dangerous incidents, https://www.newsbytesapp.com/news/science/west-midlands-ai-camera-trial-logged-nearly-19000-dangerous-incidents/tldr
- Regit: AI Cameras Catching Drink Drivers in Britain: World-First Trial Underway in Two Counties, https://www.regit.cars/car-news/ai-cameras-catching-drink-drivers-in-britain-world-first-trial-underway-in-two-counties-heres-how-they-work
- Regit: New AI cameras could help catch drunk and drugged drivers, https://www.regit.cars/van-news/new-ai-cameras-could-help-catch-drunk-and-drugged-drivers
- Highways News: Acusensus AI camera tech highlighted on national TV, https://highways-news.com/acusensus-ai-camera-tech-highlighted-on-national-tv/
- Highways News: Acusensus boss says initial automated impaired driving identification trial encouraging, https://highways-news.com/acusensus-boss-says-initial-automated-impaired-driving-identification-trial-encouraging/
- JMW Solicitors: How Do the Police Prove You Were Using a Mobile Phone While Driving?, https://www.jmw.co.uk/blog/using-mobile-phones/how-do-police-prove-using-mobile-phone-driving
- ManyAuto: The drink-drive limit this Christmas: the rules now and the change being proposed, https://manyauto.co.uk/blog/drink-drive-limit-christmas-2026.html