‘Smart’ Tube station technology that uses artificial intelligence (AI) to detect fare dodgers and passengers leaning over the edge of a platform has been quietly trialled in north west London.

Transport for London’s ‘smart’ station was tested in Willesden Green, with CCTV cameras linked to AI algorithms identifying and warning staff about concerning behaviour.

Willesden Green station was selected, as the aim was to test the technology at a local station with high footfall and minimum staffing levels.

A Freedom of Information (FOI) request revealed that it would detect patterns of behaviour, including someone jumping over a closed gate, leaning over the platform edge, or in need of mobility assistance.

Station staff would then be notified in case they needed to respond to the situation quickly.

The FoI stated: “The Willesden Green Station trial, while initially designed to alert front line teams of possible safety incidents, has provided insight into fare evasion which is building upon our current understanding and will assist us in our future approaches and interventions.”

According to the data published, after a trial between October 2022 and September 2023, the system detected 44,000 triggers and on average staff would receive 55 alerts a day.

And 720 of these alerts were related to anti-social behaviour, 66 of which were for arms raised or acts of aggression and six of them were for weapons detected.

Many of the anti-social behaviour alerts, 644, were related to unfolded bikes and e-scooters. Only four were related to smoking and vaping.

Around 60 per cent of alerts were acknowledged by station staff and only one per cent was deemed ‘invalid’ – in which the system incorrectly detected alarming behaviour.

More than 300 alerts resulted in the supervisor making a public address announcement to remind customers to stand behind the yellow line.

TfL says it is currently considering the “design and scope” of a second phase of the trial.