Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Clear out junk files and repair common Windows errors3Fix the driver behind crashes, sound loss and screen glitchesiTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Britons are encountering AI, but the latest YouGov polling suggests that exposure has not translated into broad trust. People are especially reluctant to let AI act independently or make decisions involving money, health, personal data or politics. The polls measure trust, concern and willingness to delegate—not whether everyone dislikes every AI product or company.
What the latest YouGov polling says about trust in AI
YouGov’s December 2025 polling found that trust in AI had not improved for most UK adults. In a survey of 1,287 UK adults conducted online on 4 December 2025, 20% said their trust in AI had increased over the previous year, 24% said it had decreased and 50% said it had stayed the same. The results were weighted by age, gender, education, region and social grade to represent UK adults. These are people’s reported views, not a measure of their actual use of AI.
Trust also varied by sector, but none of the tested areas had more people trusting AI than distrusting it. Technology came closest to a positive balance, while financial services and health drew greater distrust. YouGov’s December 2025 polling reported:
| Sector | Trust AI | Distrust AI |
|---|---|---|
| Technology | 28% | 38% |
| Financial services | 15% | 53% |
| Health | 18% | 48% |
The figures describe trust in AI by sector, not a universal rating of named AI companies. They also do not show that people reject every AI tool: a person may accept AI assistance in one setting while distrusting it in another.
#1 Best Overall
Where people draw the line: independent and high-stakes decisions
YouGov’s action-related questions, asked of 1,187 UK adults online on 8 December 2025, show a clear preference for human control. Respondents were asked which tasks they would never trust AI to do. The results point to boundaries around autonomy as well as around the stakes of a decision:
- Act on their behalf without approval for each action: 72% would never trust AI to do this.
- Make political decisions: 66% would never trust AI to do this.
- Make financial decisions: 64% would never trust AI to do this.
- Handle sensitive personal data: 63% would never trust AI to do this.
- Operate a vehicle without human oversight: 62% would never trust AI to do this.
Only 8% said there was no task they would never trust AI with. The pattern suggests that many people are more comfortable with AI as an assistant than as an unsupervised decision-maker. It does not establish how respondents would react to every specific product or to AI used with human review.
Rank #2
Concern about AI’s long-term risks
In the same YouGov polling, 75% of UK adults said they were concerned that AI could eventually pose a threat to humanity: 29% were very concerned and 45% somewhat concerned. The reported rounded percentages for the two levels of concern add to 74%, while YouGov reports 75% concerned overall. This is a measure of stated concern about a possible future risk, not a prediction that such an outcome will occur.
Do people want disclosure and stronger AI rules?
Earlier YouGov research on AI in media found that 86% of Britons considered it important to disclose AI use. Seven in ten also said there were not enough regulations for AI-generated content and news. Those responses concern disclosure and regulation in media; they should not be read as a direct measure of attitudes toward every use of AI. YouGov’s report, published 27 February 2025, provides that context.
The UK Department for Science, Innovation and Technology’s Public Engagement Survey 2025/2026 points to related concerns about institutions and safeguards. Only 6% of adults said they had a great deal of trust in big technology companies to keep personal data safe, and 29% said they had a fair amount of trust. Banks, financial institutions and the NHS were more trusted for data safety. At least two-thirds of adults said legislation was needed to address each tested AI risk: replacing jobs, transparency, bias and safety. These findings concern trust in companies to protect data and support for legislation; they are not a direct score of trust in AI products. See the DSIT Public Engagement Survey 2025/2026.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Are younger people rejecting generative AI?
A 12 June 2026 TechRadar report of YouGov data said 42% of British adults would eliminate generative AI entirely if they could, while 22% were undecided. Among 18–24-year-olds, 55% said they would make it disappear and 27% would not. Because this figure is reported by TechRadar as secondary coverage of YouGov data, it should be treated as a reported poll result, not as a direct measure of how often people use AI or how much they trust a particular company. The question is also about generative AI specifically, rather than AI in every form. TechRadar’s report gives the age breakdown.
High stated use and low confidence are not contradictory. Someone may use a generative AI tool for a bounded task while opposing its wider spread or refusing to let it make consequential choices. The survey figures establish attitudes, not the reasons behind each respondent’s answer.
Quick Recap
Best Value
What the polls do—and do not—show
- They show a trust gap, not universal dislike. Trust, distrust, concern and willingness to delegate are different measures. The cited polling does not provide a single net-favorability score for all AI companies.
- “AI” is not one object. General AI, generative AI, sector-specific systems and the companies behind them are not interchangeable. A response to one survey question cannot automatically be applied to all of them.
- They record stated views. These percentages are survey responses, not observed behavior or proof of how every respondent acts in practice.
- Survey wording and dates matter. YouGov’s trust measure, action-related questions, earlier media poll and TechRadar’s reported generative-AI question cover different topics and fieldwork. Compare them as related signals, not as if they were one survey.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
