British Gas to Cut 1,300 Call Centre Jobs as AI Tools Take Root in Customer Service

Centrica, the parent company of British Gas, is cutting 1,300 call centre jobs across six UK cities — Glasgow, Edinburgh, Cardiff, Leicester, Stockport, and Leeds. Of those, 800 are linked to what the company calls a "targeted deployment of AI tools," with a further 500 confirmed in the prior month. The reductions, reported on July 23, 2026, amount to a 14% cut to customer service headcount and will roll out over two years. Some roles will go unfilled after resignations; the rest will come through redundancies.
CEO Chris O'Shea framed the restructuring as a response to changing customer habits rather than a direct consequence of automation. He said over 90% of customers now start with digital channels, and the company has seen a 20% drop in customer calls. "AI isn't driving these particular job reductions; that's mainly due to changing customer behaviour," O'Shea said, adding that Centrica expects to grow jobs around its digital interface.
Trade unions have pushed back on that framing. The GMB union said 500 roles at Centrica are being replaced by artificial intelligence directly, while broader union commentary warned that the company's investment in AI would hand "hundreds of human jobs" to chatbots. The gap between what executives and unions say about the same 1,300-job reduction shows how contested the story around AI and job losses has become.
The cuts come against a mixed financial picture for Centrica's retail division. Retail profits rose to £346 million in the first half of 2026, up from £338 million in the same period a year earlier. British Gas domestic customer numbers, however, fell to 7.45 million from 7.5 million at the end of the prior year. Separately, British Gas is expected to pay up to £112 million in compensation to customers who had prepayment meters force-fitted during the Russian gas crisis.
There is a tension worth noting between O'Shea's claim that AI is not the driver and the company's own language linking 800 roles to "AI tool deployment." The distinction Centrica appears to be drawing is between AI as something that actively replaces workers and AI as a tool that handles customer queries, thereby reducing call volumes and the need for as many human staff. O'Shea places the cause on customer behaviour; the union places it on the technology.
The broader context here is the accelerating spread of generative AI and chatbot systems into large-scale customer service operations across the UK utility sector and beyond. Centrica is not the first major employer to announce reductions framed around digital transformation, but the scale — 14% of customer service teams — and the explicit link to AI tools make this a notable case for anyone tracking how AI adoption affects jobs in regulated, consumer-facing industries. The two-year implementation window suggests a managed transition rather than an abrupt displacement, which may reflect both operational logistics and sensitivity to reputational risk.
The financial picture adds another layer. A company generating £346 million in half-year retail profits while cutting frontline service jobs — and facing a £112 million compensation bill over prepayment meter practices — invites scrutiny of the relationship between cost reduction, automation investment, and customer outcomes. The slight decline in domestic customer numbers, while modest, indicates Centrica is managing retention pressures at the same time it reduces the human workforce available to handle customer queries.
For industry watchers, the key questions are whether the 20% call reduction O'Shea cites holds or accelerates, whether AI-handled customer service maintains satisfaction levels, and whether the union prediction of broader chatbot displacement proves accurate. The two-year timeline provides a measurable window. Centrica's outcomes will be watched not just by competitors weighing similar investments, but by regulators and policymakers concerned about the pace of AI-driven structural change in essential services.


