- More than five million Brits have been harmed by wrong advice from an AI chatbot – including financial losses, health scares and damaged relationships [1]
- Almost half (48%) of those who use AI for personal advice trust it as much as – or more than – a friend or family member [2]
- Nearly ten million people (18%) have made a personal decision on a chatbot’s say-so alone, with no second opinion – rising to two in five (40%) of Gen Z [3]
- One in six (17%) have told a chatbot something they have never told another person [4]
- The findings come as the UK moves to bring chatbots under the Online Safety Act
- Ernest Doku, Uswitch technology expert, shares his tips for getting the best out of the bots – and when to get a human second opinion
More than five million Brits have been harmed by wrong advice from an AI chatbot – from lost money to health scares and damaged relationships – reveals new research by Uswitch.com, the comparison and switching service.[1]
The way Britain uses the internet is shifting again – from searching the web to confiding in it. More than two in five (43%) UK adults have asked a chatbot for advice on something personal – most often their health, money or career[6] – and almost half who have used it for this purpose (48%) trust its advice as much as, or more than, a friend or family member.[2]
But the bot can be a confident bluffer. Nearly a quarter (23%) of AI users say incorrect chatbot advice has caused them a real problem[1] – ranging from financial losses (25%) to health scares (24%), damaged relationships (21%), and embarrassment (30%).
Blind faith is widespread: almost one in five (18%) UK adults – nearly ten million people – have acted on a chatbot’s personal advice with no second opinion, climbing to two in five (40%) of Gen Z.[3] Regulators are moving, with the UK Government bringing chatbots within the Online Safety Act and new EU rules requiring chatbots to remind users they are machines. But no rulebook can stop a bot being confidently wrong – so the safety net, for now, is a second opinion.
The chatbot has also become the nation’s confessional. One in six (17%) Brits have told AI something they have never told another person – doubling to 35% of under-35s [4]. The reasons are clear: it is always available (22%), private and anonymous (20%), it won’t judge (18%), and it won’t gossip (14%) – while more than one in ten (12%) simply don’t want to burden the people in their lives [7].
Used carefully, the bot can be a way in, rather than a way out. For the one in five (21%) who would feel embarrassed or judged asking a professional, a chatbot can be a first step to getting help[8] – and users are more likely to say they now open up to friends and family more since turning to AI (21%) than less (16%)[9].
Millions would even take the bot’s advice over a qualified professional[5] – drawn by answers that are faster than waiting for an appointment (43%) and cheaper than paying for advice (41%).[8]
Table: Brits who would consider an AI chatbot’s advice rather than turning to this professional [5]
| Professional | Would definitely or maybe use AI instead |
|---|---|
| A travel agent, for booking or planning a trip | 44% |
| A personal trainer or nutritionist, for diet or fitness advice | 44% |
| A career adviser or recruiter, for a job decision | 41% |
| A solicitor, for a legal question | 39% |
| A financial adviser, for a money decision | 38% |
| A GP, for a health concern | 38% |
| An accountant, for tax or money matters | 37% |
| A therapist or counsellor, for emotional support | 36% |
| An insurance broker, for cover or a claim | 34% |
| An estate agent, for buying or selling a home | 34% |
Ernest Doku’s tips for getting the best from AI advice:
- Use it as a sounding board – AI is a great way to organise your thoughts, rehearse a difficult conversation or work out what questions to ask a professional.
- Treat answers as a starting point, not the final word – AI sounds just as confident when it is wrong as when it is right, so never make a big decision on its say-so alone.
- Ask the chatbot for its sources – and check they exist. If it can’t say where its advice comes from, don’t act on it.
- Check the date – AI can rely on out-of-date information, so double-check anything time-sensitive like prices, deals, rules or rates.
- Watch what you share – never give a chatbot your bank details, passwords or anything you wouldn’t email to a stranger.
- Health concerns – check with your GP, a pharmacist or NHS 111 before acting on anything a chatbot tells you.
- Money and legal decisions – verify with free and impartial Government sites like MoneyHelper, a regulated financial adviser, solicitor or Citizens Advice, before committing to anything.
Ernest Doku, Uswitch technology expert, says: “We’ve watched Britain’s internet habits change for 25 years, and this is the fastest shift yet. We used to type a question into a search bar and weigh up the results – now millions of us ask a chatbot and take its word for it, on some of the most personal things in our lives.
“Used well, AI is a brilliant sounding board – always awake, never judgemental, and often the first step for people who would struggle to ask anyone at all. But its greatest strength – that it always has an answer – is also its biggest risk. It sounds just as confident when it’s wrong as when it’s right, and this research shows the cost when that trust is misplaced.
“So keep talking to the bot, but treat its answers as a starting point rather than the final word. If the stakes are big – your health, your money or a big life decision – get a human second opinion before you act.”
For more expert advice on the tech in your life – and the best deals to power it – visit Uswitch.com.
-ENDS-
For more information
Harriet Atkinson | Telecoms PR Manager
harriet.atkinson@rvu.co.uk
Twitter: @UswitchPR
Notes to editors
Research conducted online by Opinium, 17–21 July 2026, among 2,000 UK adults, weighted to be nationally representative. Population figures are based on the ONS UK adult (18+) population of approximately 55 million.
1. “Has an AI chatbot ever given you advice that turned out to be wrong and caused a real problem?” Base: 837 UK adults who have turned to AI for advice. Yes: 23% – equivalent to 9.8% of all UK adults, or approximately 5.4 million people. No: 64%. Not sure: 13%. Of those who said yes, the problem was: embarrassment or upset (30%), a financial loss (25%, equivalent to approximately 1.3 million people), a health scare (24%), a damaged relationship (21%).
2. “Would you say you trust an AI chatbot’s advice on a personal issue more or less than advice from a friend or family member?” Base: 837 UK adults who have turned to AI for advice. Much more (4%), slightly more (16%), about the same (29%), slightly less (25%), much less (20%), not sure (7%). NET trust AI more or the same: 48%.
3. “Considering your use of AI, have you ever acted on personal advice from an AI chatbot without checking it with anyone else first?” Yes: 18% of all UK adults, equivalent to approximately 10 million people; 37% of 18–34s; 40% of Gen Z (18–29); 21% of men vs 15% of women.
4. “Considering your use of AI, have you ever told an AI chatbot something personal that you have never told anyone else in your life?” Yes: 17% of all UK adults, equivalent to approximately 9.5 million people; 35% of 18–34s.
5. “For each of the following, would you use an AI chatbot’s advice for a personal need / issue rather than turning to this type of professional?” Figures show “Yes – definitely would” plus “Maybe – it would depend on the situation”: travel agent (44%), personal trainer or nutritionist (44%), career adviser or recruiter (41%), solicitor (39%), financial adviser (38%), GP (38%), accountant (37%), therapist or counsellor (36%), insurance broker (34%), estate agent (34%).
6. “Have you turned to an AI chatbot for advice on any of the following?” 43% selected at least one topic. Physical health advice (16%), financial advice (15%), travel/holiday advice (14%), career advice (12%), mental health advice (11%), family issues (9%), legal advice (8%), relationship or dating advice (8%). Other (1%).
7. “What, if anything, would make you more likely to confide in an AI chatbot rather than a person?” It’s available any time, day or night (22%); it’s private and anonymous (20%); it’s free, or cheaper than professional advice (18%); it won’t judge me (18%); it will offer impartial advice (14%); it won’t gossip or tell anyone else (14%); it doesn’t get bored or tired of me (12%); I don’t want to burden the people in my life (12%).
8. “You mentioned you would consider using an AI chatbot’s advice over a professional’s in some situations. Why is that?” Base: 1,216 UK adults who would consider AI over at least one professional. It’s faster than waiting for an appointment (43%); it’s available any time, day or night (42%); it’s free, or cheaper than paying a professional (41%); the advice can be easily verified (29%); I don’t want to feel embarrassed or judged asking a professional (21%); I don’t trust professionals to always have my best interests at heart (18%).
9. “Since you started using AI for personal advice, has the amount you confide in friends or family changed?” Base: 837 UK adults who have turned to AI for advice. Yes, a lot more (6%); yes, a little more (15%); no change (59%); yes, a little less (13%); yes, a lot less (3%). NET confide more: 21%; NET confide less: 16%.