Geoff Cross, managing director of AI-powered vehicle CCTV and telematics firm Centrad, looks at why the next decade of road safety could be defined by how effectively we use technology to address human error.

Recent government statistics paint a mixed picture of the current state of play of road safety across the UK. Data shows that while deaths are down, serious injuries are on the rise. And this highlights the need for a different approach to reducing collisions.

In 2025, an estimated 1,556 people lost their lives in road collisions, a 3% decrease compared with the previous year. While that might suggest progress at a first glance, the broader picture tells a different story.

According to the data, the number of people killed or seriously injured increased by 4% year on year to 29,911. Looking beyond the most recent data, over the past five years the picture has remained largely unchanged.

Although fatalities have gradually declined, the number of people killed or seriously injured has remained stubbornly close to the 30,000 mark each year.

To make a tangible change on road casualty statistics over the next decade, achieving meaningful improvements in road safety will require a complete shift in focus.

Despite investment into infrastructure upgrades, safety initiatives and new policies, the fact that incident numbers remain stagnant shows that there is still work to be done.

Improving road safety requires more than changes to roads or legislation. It also means addressing the human factors that contribute to collisions.

People will always be at the centre of road safety, but expecting drivers to make the right decision every second of every journey is unrealistic. As a former fleet operator and driver, human error is an unavoidable part of driving, and always will be, which means the next major improvements need to focus on reducing the risks associated with human behaviour.

The Royal Society for the Prevention of Accidents states that around 90% of road crashes involve some element of human error.

Today, many fleet safety technologies and processes remain reactive. A collision occurs, footage is reviewed, and action is only taken afterwards through measures such as additional training, internal policy changes or driver intervention.

Rather than simply recording incidents after they occur, AI can identify patterns of risky behaviour, enabling operators to intervene before a collision happens.

The real value in the years ahead won’t be the hardware itself. The differentiator will be the software, machine learning and the insights generated from the data.

AI-connected camera systems are already changing the way fleets approach safety. Rather than capturing footage after an incident, these systems can detect warning signs such as distraction, fatigue, mobile phone use or unsafe manoeuvres as they happen, delivering immediate alerts to drivers and fleet managers to help prevent collisions.

The real opportunity for AI lies in what it could do next. Future systems could combine driver behaviour, working patterns and circadian rhythms to predict fatigue before it becomes a safety risk, allowing fleets to intervene earlier. AI could also move beyond driver behaviour, with connected cameras identifying risks such as vandalism and anti-social behaviour. The technology will evolve from simply recording incidents to both predicting and preventing them.

Looking to the future, driving down accident statistics will hinge on this ability to move from a reactive approach to a preventative one.

With the UK government targeting a 65% reduction in people killed or seriously injured on British roads by 2035, the industry needs to actively explore new opportunities to improve safety.

The percentage drop between 2015 and 2024 currently sits at around 17% so to achieve that goal, there’s a lot of work to be done over the next decade.

We can’t afford to rely on the same approaches that have yielded limited progress. AI technology, such as intelligent cameras, driver monitoring systems and predictive analytics, provides a real opportunity to unlock a completely new dimension to road safety – which is already in progress.

As this technology develops and evolves, machine learning will become increasingly capable of identifying patterns across millions of journeys, recognising behaviours and circumstances that may increase the likelihood of an incident. But that won’t translate to AI replacing the need for skilled drivers, effective training or safer infrastructure.

That will always be crucial to any successful road safety strategy. For years, much of the technology used in fleets has helped us understand what happened after an incident. The real opportunity with AI is to move beyond that and identify the warning signs early enough to prevent the incident happening in the first place.