42% of UK adults are deliberately limiting their use of AI

The top reason is privacy and security, not confusion. Just 6% of those cutting back say they feel unprepared to use the technology.

Noodle News data desk · reading time ~4 min

42%

of UK adults deliberately limit how much AI they use¹

70%

think it would be difficult or impossible to avoid AI exposure entirely¹

of limiters cite feeling unprepared as their reason — the least-picked option¹

What the data shows

Nearly half of Britons are pulling back — and privacy, not confusion, is why

A survey of 2,055 UK adults for King’s Digital Futures Institute and Responsible AI UK finds that 42% deliberately limit how much AI they use; 46% say they don’t, and 12% don’t know.¹ Use is common — 46% use tools such as ChatGPT, Claude, Gemini or Copilot weekly, and 32% don’t use AI at all¹ — so pulling back is mostly a choice, not a gap left by people without access.

Limiters were asked to pick the single most important reason. Privacy, security and compliance concerns led by a wide margin, ahead of a simple preference to keep working the way they already do.¹

Exhibit 01

Why people who limit their AI use say they do it

Source: Deltapoll survey for King’s Digital Futures Institute & RAi UK, 12–15 Jun 2026, n=2,055 UK adults; reasons question base is the 42% who report limiting use (n≈863). Single most-important reason, forced choice.¹

Key finding. Only 6% of people who limit their AI use say it’s because they don’t feel prepared to use it effectively — the least-picked reason on the list. This isn’t a skills gap; it’s a deliberate choice.¹

The Gen Z paradox

The people most confident with AI are also the most likely to limit it

Gen Z use AI more regularly than Millennials or Boomers, and 93% call themselves “very” or “fairly” confident using it — yet they are the age group most likely to limit their use.¹ Confidence and comfort aren’t the same thing: even among daily AI users, more than a third still say it carries more risk than benefit.¹

The pattern holds economy-wide. Between October 2023 and June 2026, the share of adults who see more risk than benefit rose 4 points, 48% to 52%, while those seeing more benefit than risk fell 4 points, 38% to 34%.¹ Belief in personal benefit moved both ways at once — agreement rose 6 points (32% to 38%) but disagreement rose faster, 8 points (19% to 27%) — a drift away from the middle, not a swing to one side.¹

What we don't know

This is the survey’s initial data release; the full academic paper and open dataset are still pending, so the 42% figure has not yet been independently replicated by a second UK survey using the same question. “Limiting” use is self-reported and self-defined — the survey doesn’t measure actual usage logs, and given 70% of adults think avoiding AI exposure entirely would be difficult or impossible, some self-reported “limiters” may still be using AI-embedded tools without registering it as AI use. The reasons question forced a single “most important” choice, which likely understates how often privacy and habit act together. The Oct 2023 comparison point comes from a differently-commissioned survey (the government’s Centre for Data Ethics and Innovation) using the same pollster and question wording, which supports the comparison but isn’t a like-for-like replication.¹

Caveat. The strongest objection to reading this as a settled "AI backlash": self-reported avoidance intentions can diverge from actual behaviour, and a single-wave survey can’t show whether today’s 42% is a stable position or a snapshot mid-swing.

Go deeper

Demographics. The report breaks attitudes down by gender (men report more optimism about AI benefits than women) and by income (higher earners use AI more often and see more benefit) — not covered here for length.¹
The politics of pushback. The report situates this survey against UK policy flashpoints — the government dropping its AI copyright “opt-out” proposal, and public petitions against AI-driven contracts — that this piece doesn’t explore.¹

Method. All percentages are as reported in the source survey; no figures in this piece are modelled, projected or estimated. The Exhibit 01 "all other reasons" bar aggregates seven smaller categories (not permitted at work 5%, no workplace access 3%, no time to learn 3%, not useful 2%, cost 1%, other 1%, don't know 3%); the aggregation and all percentage-point deltas in the text were computed and checked in Node.js against the source table before publication. Margin of error (95% CI, unweighted approximation) is ±2.1 points on the 42% headline figure (full sample, n=2,055) and ±3.0 points on the reasons breakdown (limiter sub-sample, n≈863); the underlying data are weighted by the pollster to be nationally representative, so true error bands will differ slightly from this approximation.

Sources. Numbered, highest tier first; each inline citation points here.

T1Devlin, K., Stilgoe, J., Orchard, D. & Peter, O. AI: the growing UK pushback — research briefing and initial findings. King’s Digital Futures Institute & Responsible AI UK, 10 Jul 2026. Survey by Deltapoll, fieldwork 12–15 Jun 2026, n=2,055 UK adults, weighted to be nationally representative. rai.ac.uk/reports/ai-the-growing-uk-pushback (DOI 10.18742/pub01-262), retrieved 26 Jul 2026.

Corrections. None. Material corrections will be logged here with date.

Figures to 27 Jul 2026. Noodle News · v1.0 · uk-ai-avoidance.

42% of UK adults are deliberately limiting their use of AI

Nearly half of Britons are pulling back — and privacy, not confusion, is why

The people most confident with AI are also the most likely to limit it

Read more