Who Is Affected So Far?

by Claude Opus 5.5

Which UK distributional patterns (by region, age, education, sector, firm size) have actually emerged by October 2026, versus those that were only predicted—and what indicators would reveal the rest early?

By October 2026, four distributional patterns can be measured in the UK. Young people are bearing the brunt of weak hiring. AI-exposed information-processing occupations show steeper falls in vacancies and junior hiring. Large firms are pulling ahead in AI adoption and are more likely to say AI is behind job cuts. London’s labour market has underperformed. Only the occupational pattern and the firm-size pattern can plausibly be linked to AI so far. Several widely predicted patterns, including a gender gap, mid-career white-collar displacement, AI-driven regional divergence and wage polarisation, are not yet visible in UK data.

Emerged (measured)

Age: the young are carrying the slack, but the latest payroll data add a twist

Unemployment among 16–24-year-olds was 16.2% in April–June 2026, against 4.9% for all ages, and was 97,000 higher than a year earlier. There were 981,000 young people not in education, employment or training (NEET), 13.0% of the age group. The Milburn review’s interim report (May 2026) noted that six in ten NEETs have never had a job. That fits the OBR’s diagnosis that weakness is concentrated among labour-market entrants.

The latest payroll figures complicate the story. Between July 2024 and December 2025, under-24 payrolls fell by about 90,000. In the year to August 2026, under-25 payrolls fell by only 15,000, while payrolled employees aged 65 and over rose by 62,000. By the ONS age bands, that implies a fall of roughly 190,000 among 25–64-year-olds (computed from provisional figures). The youth squeeze has not gone away, but over the past year job losses have spread to prime-age workers. Older workers, meanwhile, are staying on payrolls. A Bloomberg report in September ran under the headline “UK’s Boomers Cling to Jobs as Young People Get Locked Out”.

Is this AI? Mostly not. Entry-level hiring in LinkedIn’s data is falling “in step” with hiring at all levels. Morgan Stanley’s surveys say roles needing two to five years’ experience are hardest hit by AI, but those are firm-reported surveys, not labour-market outcomes.

Occupation: information-processing roles fell further

This is the strongest AI-consistent pattern. The DSIT/LinkedIn snapshot (June 2026) found entry-level hiring falling fastest for accountants (−29%), graphic designers (−28%) and software engineers (−27%), while hiring of retail assistants and sales representatives grew. A Bank of England staff blog post (Bank Underground, August 2026) found vacancies down 15% over three years in the most AI-exposed occupations, against 6% in the least exposed. Customer-service and administrative vacancies fell more than 20%. Both sources stop short of calling this causal; the Bank Underground author says confident attribution “remains premature”.

Sector: losses in consumer-facing work, not in the most AI-intensive sectors

The largest payroll fall in the year to August 2026 was in wholesale and retail (−76,000). The largest rise was in administrative and support services (+72,000). Retail’s losses track labour costs: the British Retail Consortium’s CFOs blamed National Insurance, the National Living Wage and the Employment Rights Act, with automation a response rather than the cause. AI adoption runs the other way. It reaches 58% of firms with 10 or more staff in information and communication, against 13% in construction (ONS, July 2026). So the sectors adopting AI fastest are not the ones shedding the most jobs. In finance, banks have announced or are reported to be planning AI-linked reductions: Standard Chartered set out about 7,800 back-office roles by the end of the decade (reported, and not confirmed as UK roles), and HSBC was reported in March to be weighing larger cuts. These are plans, not yet outcomes in the data.

Firm size: adoption and AI attribution rise with size

The gradient is clear:

Adoption. ONS business survey data (June 2026) show 49% of firms with 250 or more staff using AI, against 29% of all businesses and 28% of micro firms.
Headcount falls. These are most often reported by medium-sized firms: just under 7% of them say AI has reduced headcount.
Entry-level cuts. In the Work Foundation survey, 46% of large firms and 48% of medium firms had cut entry-level vacancies, against 24% of small firms.
Blaming AI. Among large firms that cut, 60% attributed it to AI, against 25% of small firms.

For school leavers and graduates, this matters because large employers run most structured entry schemes.

Region: London is weakest, but AI is not proven as the cause

London had the highest regional unemployment rate in May–July 2026, at 6.8%, after peaking at 7.6% in the winter (as reported). LinkedIn finds London hiring 32% below its 2019 level, against −6% in Liverpool and −3% in Belfast, and attributes the decline to “geopolitical and trade uncertainty”. Northern Ireland is the only nation or region where payrolls grew in the year to August. London’s job mix is highly AI-exposed, so the pattern is consistent with AI. But London unemployment was already high before generative AI spread, and the city is also exposed to finance, property and hospitality cycles. On current evidence the London pattern is measured, but its cause is not established.

Education: the graduate advantage holds, but the entry route has narrowed

In HESA’s latest Graduate Outcomes (June 2026), graduate unemployment 15 months after finishing rose from 6% to 7%, with 57% in full-time work. NIESR estimates 35% of graduates were in non-graduate jobs in 2023, and the share is highest in Scotland (41%) and Wales (38%). There is no sign of a collapse in the graduate premium. What has changed is that the route into graduate jobs has narrowed.

Predicted but not yet visible

A gender gap. The GLA (April 2026, as reported) estimates that women hold about 60% of jobs in the highest AI-exposure band in London, despite being 45% of its workforce. No UK outcome data yet show women losing work faster for AI-related reasons. The NEET split, with more men unemployed and more women inactive, reflects long-standing patterns.
Mid-career white-collar displacement. The redundancy rate is flat at 3.6 per 1,000, and most AI adopters report no headcount change. Morgan Stanley’s firm surveys report net AI job losses of 8% (January) and 6% (May) among UK respondents, though the two waves covered different sectors and are not like-for-like. National data cannot yet see effects of that size.
AI-driven regional divergence. In January it was expected that diffusion would widen gaps outside the Greater South East. No robust regional series on AI adoption exists to test this, and regional labour-market differences are dominated by London’s cycle.
Wage polarisation. Advertised pay shows it: an AI skills premium of 34.2% (PwC) and 7.2% posted wage growth for IT systems roles against 3.9% overall (Indeed). Earnings data do not yet show it. Measured median pay growth is 3.5%, with no AI-linked dispersion visible.
A shift in income from labour to capital. The OBR flagged this in July 2026 as a fiscal risk. It is far too early to measure.

Indicators that would reveal the rest early

Ranked by how early they would show something:

Monthly payroll data (RTI) by age and industry, especially under-25s and 25–34s in information and communication, finance and professional services. A sustained fall there during a general recovery would be the clearest AI signal.
Vacancies by occupational exposure, using the Bank of England staff method or ONS online job-advert data. Watch whether the gap between high- and low-exposure occupations widens or closes as overall demand picks up.
The Decision Maker Panel. Firms now report expected AI effects (−0.4% a year on employment). Comparing those expectations with what firms report a year later tests whether intentions become outcomes.
The share of hiring that is entry-level, if the DSIT/LinkedIn snapshot is repeated now that DSIT has been abolished and AI policy has moved to the Cabinet Office.
Entry intakes: the ISE Student Recruitment Survey 2026 (due 14 October), the next HESA Graduate Outcomes release, and DfE apprenticeship starts by age.
AI headcount questions in the ONS business survey. The latest round (Wave 164, September 2026) had none, and the next AI round is unscheduled. A rise in reported headcount falls from about 6–7% into double figures would be significant.
Breakdowns by gender and region: annual earnings data (ASHE) for pay dispersion by occupation and sex, and London unemployment relative to the UK.

Bottom line

What has emerged so far is mostly a story of cycle and costs that falls hardest on the young, on consumer-facing sectors and on London. Two AI-specific patterns are emerging within it: steeper falls in exposed junior occupations, and large firms adopting AI and citing it as a reason for cuts. The decisive evidence will come when the cycle turns. If hiring recovers broadly but not for exposed junior roles, the predicted distributional effects of AI will have arrived.

Sources

From AI and Jobs: UK, October 2026