61% of UK financial services staff now use generative AI daily, Smarsh finds

TL;DR:

• Smarsh’s 2,000-person survey of UK financial services and insurance professionals finds 61% use generative AI every day, with the technology embedded in 50% of reporting tasks, 49% of call notes, 40% of client communications and 34% of compliance documentation.
• Only 41% thoroughly review and edit AI-generated material, and just 32% believe their firm’s surveillance systems can detect AI-content risks.
• The findings give UK regulators (FCA, PRA, ICO) the most concrete picture yet of how deeply generative AI has entered regulated communications — and reframe board-level conversations about supervision and audit, rather than adoption.

Generative AI use is no longer a pilot phenomenon inside UK financial services. New research published by communications-surveillance vendor Smarsh, based on a survey of 2,000 UK financial services and insurance professionals, finds 61% are using generative AI every day. The data shows the technology is now embedded in 50% of reporting tasks, 49% of call notes and summaries, 40% of client and customer communications, 38% of marketing and social-media content, 37% of internal communications and 34% of compliance documentation.

The figures matter for one specific reason: only 41% of respondents say they “thoroughly review and make significant edits” to AI-generated material before it goes out the door. The survey shows AI is deeply embedded in regulated outputs without proportionate human oversight, in a sector where the FCA and PRA expect firms to retain records, supervise communications and demonstrate that customer-facing materials meet regulatory standards.

Where the oversight gap sits

Confidence in surveillance is low. Just 32% of respondents say their organisation’s surveillance systems are “fully equipped” to detect risks in AI-generated content. The under-confidence is most pronounced among the heaviest users: 43% of 25 to 34-year-olds — more than 36% of whom report using AI multiple times a day — see surveillance gaps. Daily use is also widespread among older cohorts, with nearly a third of those aged 35 to 54 and 28% of those 55-64 reporting daily AI use, suggesting this is not a generational bubble.

Notably, staff frame stronger oversight as enabling rather than restricting use. Across the sample, 81% say they would feel more confident using AI for work if they knew the firm was monitoring outputs — rising to 87% among 18 to 34-year-olds. That figure was 12% higher than when the same question was asked a year earlier, indicating users are catching up with the compliance reality before their firms do.

Why this matters in 2026

The numbers slot into a wider UK compliance picture: the FCA’s ongoing engagement with firms on generative AI use, the PRA’s interest in operational resilience, and the ICO’s existing requirements on automated decision-making. The Smarsh data effectively gives regulators a concrete adoption baseline: roughly six in ten UK finance and insurance workers using generative AI daily across regulated communications, with self-reported supervision gaps that mirror EU concerns under MiCA and DORA. Some 69% of respondents say AI is increasing the volume of content they create — meaning the supervisory surface is also expanding faster than headcount.

Smarsh has a commercial interest in the framing — the company sells communications surveillance — but the underlying numbers are consistent with anecdotal patterns reported across the City. Paul Taylor, Smarsh’s vice president of product, argues “compliance leaders are now under pressure to ensure every AI-assisted interaction is transparent, supervised, and defensible”. For UK firms, that translates into specific control questions: which workflows allow AI drafting? Which require sign-off? How are AI-generated calls and summaries retained for the same period as human-authored equivalents?

Looking forward

Expect UK firms to move from “AI policy” to “AI surveillance and audit” as the next investment frontier — and expect compliance budgets to find new lines for tooling, prompt logging and AI-content retention. UK boards that have not yet asked their second line of defence whether monitoring keeps pace with the 61% daily-use number now have a directly cited industry data point to do so.