Force uses AI to cut wait times on 101 calls
South Yorkshire Police has become the first force to fully roll out an AI-powered call-triage system on its 101 non-emergency line, aiming to cut waiting times by weeding out misdirected, nuisance and prank calls before they reach a human handler. The government-backed technology matches a caller's stated reason for calling to a set of likely categories and redirects non-police issues, such as council or NHS matters, to the correct service before the caller joins the queue, freeing up call handlers to focus on genuine policing needs.
The force received more than 470,000 calls to 101 in the year to June 2026, with current wait times of eight to 15 seconds. Nationally, the 101 line takes around 20 million calls a year, roughly a fifth of which are meant for other organisations, alongside trivial reports such as late pizza deliveries or slow pub service. The £1.4m Natural Language Processing software, developed by the Home Office, National Police Chiefs' Council and Vodafone, was trialled in 15 forces and tested for regional accent recognition before going live in South Yorkshire on 18 August.
- South Yorkshire Police rolls out AI triage system for 101 calls
- AI redirects non-police calls before they join the queue
- £1.4m system aims to cut wait times and free up officers
New here? Start with this
Police forces use 101 for non-emergency matters, such as reporting a crime that has already happened or asking for advice. It is separate from 999, which is for emergencies. Callers may also contact 101 about issues that belong with another organisation, such as a council or the NHS.
Police call handlers deal with a wide range of calls, and time spent on misdirected or nuisance calls can add to the work waiting in the queue. The Home Office, the National Police Chiefs’ Council and Vodafone developed software that uses callers’ descriptions to sort calls into likely categories. It has been trialled by several forces, with testing to help it recognise regional accents.
South Yorkshire Police handles hundreds of thousands of 101 calls each year. The system is intended to direct callers with non-police issues to the relevant service and leave call handlers more time for policing matters.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
The system addresses a quantified, genuine problem: roughly one-fifth of the 20 million annual 101 calls are misdirected to other services or report trivial matters, clogging the line. The technology has undergone extensive trialling across 15 forces with specific validation for regional accents before deployment, meaningfully reducing implementation risks. By efficiently redirecting non-police matters to appropriate services, the system frees call handlers to focus on genuine emergencies, improving response times for those who genuinely need police assistance whilst simultaneously providing faster resolution for callers whose needs lie with councils or the NHS.
The case against
Deploying AI as a gatekeeping mechanism for emergency services carries risks that testing cannot fully eliminate. Natural language processing can misclassify complex or ambiguous situations where rapid human judgment proves essential, and callers with accents or communication patterns outside the system's training data risk wrongful redirection away from needed police assistance. The system removes human discretion from determining what constitutes a genuine police matter, creating rigid categorisation that fails to account for contextual nuance and raising accountability questions about responsibility when misclassification occurs.