Money worries, career dilemmas, relationship problems and mental health struggles are increasingly being shared not with a friend, but with a chatbot. According to new research from privacy company Proton, 66% of British AI users have discussed at least one sensitive topic with an AI assistant.
The appeal is easy to understand. A chatbot is available at 2am, doesn’t judge and won’t bring the conversation up again later. But the same research suggests that people are confiding in AI far faster than they are learning to trust the firms that run it.
What Are Brits Telling AI?
Personal finance is the sensitive topic UK respondents most often raise with a chatbot, with a third (33%) saying they had discussed it. The rest of the list reads like the contents of a late-night conversation with a close friend:
- Personal finance: 33%
- Work or career problems: 32%
- Mental health struggles: 27%
- Relationship or dating problems: 20%
- Sex life: 15%
- Family conflicts: 12%
- Secrets: 10%
The UK is not alone in putting money first. Proton’s wider study, which covered France, Germany and the US as well, found finances were also the top topic in France, while mental health led in Germany and career problems led in the US. Across all four countries, Americans were the most likely to have shared something sensitive with AI (68.4%) and the French the least likely (60.3%).
AI is also shaping behaviour beyond conversation. More than half (55%) of Brits say chatbots have changed the way they make everyday decisions.
Why A Chatbot Rather Than a Person?
The research does not suggest AI is replacing human support. Friends and family remained the preferred choice for every sensitive topic surveyed, ranking ahead of chatbots, therapists and religious advisors.
Instead, AI appears to fill a specific gap. For British users, freedom from judgement was the leading reason for choosing a chatbot over a person when discussing mental health, sexual health, secrets and family conflicts. Trust and privacy came top for conversations about romantic attraction, while convenience was the main draw for work issues.
There is a notable UK quirk here. Across the four countries, convenience was the top reason for discussing personal finances with AI everywhere except Britain. On mental health, UK and US users put freedom from judgement first, whereas French and German users cited convenience.
“AI is learning far more about us than a search engine ever could,” said Eamonn Maguire, Director of AI Engineering at Proton. “We’re not just asking questions. We’re sharing deeply personal context about our health, finances and relationships. And when technology knows that much about us, privacy can’t be an afterthought.”
A Trust Gap at the Heart of AI Use
Despite this openness, confidence in AI providers is low. 42% of UK respondents said they have little or no trust in AI companies to protect sensitive information shared with chatbots, and 11% said they don’t trust them at all.
Across all four markets, only between 11% and 20% of respondents reported a high level of trust in AI companies with their private data. Low trust was most common in France (47.7%), followed by the UK (41.6%), Germany (40.5%) and the US (38.8%).
More from News
- Video Game Companies Are Abandoning “Buy Our Console To Play This” – Here’s Why
- UK Gambling Operators Face New Deposit Limit Rules From 30 September
- How A Ransomware Gang Hijacked A Rival Gang’s Website – And Is Now Extorting Them For “2.333% Of Their Net Worth
- Meta Promised Unmatched Privacy For Muse – So How Did Researchers Crack It So Quickly?
- Digital Avatars Are Bringing Deceased Artists Back To The Stage – Where Do We Draw The Line?
- Can Europe Solve Its Critical Minerals Problem Fast Enough To Support Its Tech Ambitions?
- Google Has Agreed To A £260 Million Settlement: Who’s Eligible For Compensation?
- China Wants To Build A Third Global AI Bloc – What Does That Mean For Everyone?
Advertising Profiles Worry People More Than AI Training
Perhaps the most striking finding is what users are actually worried about. In the UK, 64% are concerned their conversations could be used to build advertising or marketing profiles, compared with 52% who are concerned about their chats being used to train AI models.
The distinction matters. Any single conversation may seem harmless, but a chatbot that hears about someone’s debts one week, a health scare the next and a relationship wobble after that could, over time, build a remarkably detailed picture of a person’s life. The concern for many users is less that an AI remembers what they said, and more that it could be turned into a commercial asset.
The gap was widest in the US, where 65.3% were worried about ad profiling against 57.6% for model training.
“People have already decided that AI is useful enough to hear their money worries, doubts and everyday concerns,” said Maguire. “They are handing AI companies their inner lives and biggest secrets on the assumption that they will protect them when it’s actually the opposite. AI’s business model depends on learning from them.”
Privacy Could Decide How Far People Go
The findings suggest privacy isn’t what gets people to try AI, but it may determine how much they are prepared to share once they do. 44% of British respondents said they would be more willing to share sensitive information if their conversations were guaranteed not to be used to train AI models, a figure that ranged from 40.7% in France to 47.7% in the US.
Interest in privacy-first tools was even stronger. 82% of UK respondents said they would be more likely to use an AI chatbot specifically designed around privacy, the highest of any country surveyed; Germany was lowest at 71.1%.
Proton, which offers its own privacy-focused assistant Lumo, has also launched AI Paper Trail, a tool designed to show people how much personal information their AI conversations may reveal about them.
For AI providers, the message is fairly clear. Users have already decided chatbots are useful enough to hear about their finances, health and relationships. Whether they keep sharing may depend on which companies can prove, rather than simply promise, that those conversations are safe.
