Legal & General to cut 1,000 jobs by mid-2027

You have probably sat in a town hall where the chief executive promised that technology would free people to do higher-value work. The script is familiar: automation handles the routine, humans focus on judgement, and the organisation emerges leaner and more capable. Legal & General, the UK insurance and investment giant, is now testing that promise at scale. The firm will cut 1,000 roles by mid-2027, a reduction driven in part by artificial intelligence taking over tasks that once filled spreadsheets and inboxes.

The announcement lands at a moment when financial services are racing to embed generative AI into underwriting, claims processing and customer service. Legal & General employs roughly 10,000 people, so a thousand departures represent a tenth of the workforce. The company has not disclosed which functions will shrink, but the pattern across the sector suggests that administrative, data-entry and first-line support roles are most exposed. Insurers have long automated actuarial calculations and risk modelling; generative AI now extends that logic to drafting policy summaries, triaging claims and answering policyholder queries in natural language.

What this means for capability

The Legal & General cuts raise a question that learning and development teams cannot dodge: if a tenth of your organisation disappears because machines can do the work faster, what capability do the remaining nine-tenths need? The reflex answer is digital literacy, but that phrase has become a catch-all that obscures more than it clarifies. Literacy implies reading and writing; what employees actually need is the ability to prompt, audit and override AI outputs when the model halts or hallucinates.

Consider a claims handler who once spent three hours reviewing documents and another two drafting a settlement letter. A large language model can now summarise the documents in seconds and generate a draft in a minute. The handler's new job is to spot when the summary omits a crucial exclusion clause or when the draft tone will inflame a grieving family. That skill is not taught in a two-hour e-learning module titled "Introduction to AI." It requires domain expertise, emotional intelligence and enough technical confidence to interrogate the machine's reasoning. Legal & General's workforce will need all three, and the firm's L&D function must design for that triad rather than for generic prompt engineering.

The redeployment challenge

Legal & General has not said whether the thousand departures will be voluntary redundancies, natural attrition or forced exits. Each path carries a different capability burden. Voluntary schemes often lose the most marketable talent first, leaving behind employees who struggle to adapt. Natural attrition is slow and uneven, creating pockets of understaffing while other teams remain overstaffed. Forced redundancies can be swift but demoralising, and they risk cutting people who could have retrained for adjacent roles if the organisation had invested early enough.

The firms that navigate this transition well treat redeployment as a design problem, not a human-resources afterthought. They map which tasks AI will assume, which new tasks will emerge and which existing employees have the aptitude to move into those roles. A customer-service agent who has spent five years resolving complaints by phone may have the empathy and product knowledge to become an AI quality auditor, reviewing chatbot transcripts for tone and accuracy. A back-office clerk who has mastered Excel macros may have the logical mindset to configure robotic process automation workflows. Both transitions require training, but they are cheaper and faster than hiring externally, and they preserve institutional memory that no onboarding programme can replace.

The L&D stake

For learning and development leaders, the Legal & General announcement is a forcing function. If your organisation is planning similar cuts, you have perhaps 18 months to build the capability infrastructure that will keep the remaining workforce productive and the departing workforce employable. That infrastructure has three layers.

First, baseline AI literacy for everyone. Every employee should understand what a large language model can and cannot do, how to write a prompt that yields a useful output and when to escalate to a human. This is not optional for "tech roles"; it is foundational for anyone who will work alongside automated systems.

Second, role-specific upskilling for people whose jobs will change but not disappear. A mortgage underwriter who once spent hours verifying income documents will now spend minutes reviewing an AI summary and more time on complex cases that require judgement. That shift demands training in risk assessment, regulatory edge cases and how to document decisions when the initial analysis came from a black box.

Third, transition support for people whose roles will vanish. Outplacement services and résumé workshops are table stakes; forward-looking firms also fund short courses in adjacent skills and create alumni networks that connect former employees to contract opportunities. Legal & General has not detailed its support package, but the reputational cost of a poorly managed redundancy programme can outlast the cost savings.

What comes next

The Legal & General cuts will not be the last. Insurers, banks and asset managers are all running pilots that replace human effort with machine inference. The question for L&D is whether your organisation will treat this as a one-time restructuring or as the first iteration of a continuous capability cycle. If AI improves at the pace of the past two years, the tasks it can handle in 2029 will make today's automation look modest. Organisations that build learning cultures capable of absorbing that pace will retain talent and agility. Those that treat each wave of cuts as a surprise will lose both.

The promise that technology frees people for higher-value work is not a lie, but it is conditional. It depends on whether the organisation invests in the capability that makes higher-value work possible, and whether it does so before the town hall script turns into a redundancy notice.

Sources:

• The Guardian, 23 September 2026