(Credit: © Wordley Calvo Stock - stock.adobe.com)
Distrusting the Internet Does Not Stop Adults From Using It
In A Nutshell
- Of 22 UK adults interviewed, 12 said they don’t trust the internet but kept using online tools or accepting cookie settings anyway.
- Convenience, time pressure, and a feeling of having no real choice were common reasons.
- Distrust showed up in people with and without a mental health diagnosis, though only two participants had psychosis.
- Reputation mattered: a university-developed app earned relatively high trust, while social media drew suspicion.
In a new study of 22 adults across the UK, 12 said they don’t trust the internet, yet they kept using online tools or going along with cookie settings that left them uncomfortable. One participant summed up the bind: “And I don’t trust it but as I said before, I have to in a way trust it in order to get to do what I want to do.”
Researchers interviewed adults with and without a history of mental health problems about artificial intelligence, internet-based tools, and digital health apps, often called eHealth. Trust is often treated as a requirement for using new technology, but in this group it did not settle what people actually did online.
Differences by mental health status were what the team expected to find. Instead, distrust showed up in people with and without a diagnosis. Published in the journal Behavioral Sciences, the study describes people who want privacy but feel unable to opt out of a world that runs through apps, websites, and online accounts.
Interviews With 22 UK Adults Found Distrust of Online Technology Was Common
Participants were recruited online between March and October 2023. Eight of them self-reported a mental health diagnosis (six with depression, two with psychosis, a condition that can involve intense suspicion), and 14 reported no history of mental health problems. Ages ranged from 23 to 75. Most were women. Most were also white, which the researchers say should be kept in mind when interpreting the results.
Before each interview, participants spent 20 minutes on a university-developed brain-training app, so everyone had recent hands-on experience with a simple digital health tool. Each person then took part in an interview lasting 30 to 45 minutes, and everyone chose online over in-person. Questions covered AI, cookies, and how much trust matters online. Researchers then read the transcripts for recurring ideas.
Convenience Pushes Adults to Use Online Technology They Distrust
Half of the participants rated trust as very important when using online tools, and it mattered most when sharing sensitive or private information. Yet convenience often won out. Anyone who has hurriedly accepted a cookie banner to reach the page underneath will recognize the move, and one participant admitted, “sometimes I just agree to it if I’m in a hurry.” Given the choice, participants said they preferred to reject non-essential cookies to limit the personal data collected about them.
Others described a sense of obligation. One participant compared the situation to being threatened and asked to hand over money: agreeing is technically a choice, but hardly a free one. A similar tension showed up with the brain-training app: participants wanted increasingly personalized suggestions but understood that meant sharing more data.
About two-thirds of participants described emotional reactions to AI and online tools, ranging from anxiety, frustration, and worry to more positive feelings. One participant said rejecting cookies felt designed to “wear you down.” Others found a source of confidence in their wariness. “Maybe we’re more in tune now just to spot dodgy things,” said one, using British slang for shady.
Trust in Online Technology Depends on Who Built It and Who Is Using It
Reputation mattered: participants cared about who made a technology and how that organization had behaved. Just over one-fifth voiced distrust of social media sites, with some citing the Cambridge Analytica data scandal. The brain-training app earned relatively high trust, often because of where it came from. One participant said it seemed trustworthy “because obviously it’s being done by a university.”
Comfort with technology mattered too. Older participants tended to report more anxiety about digital tools, and a participant in their 70s said, “I’m not very good technically, and that makes me apprehensive.” Participants also spoke of wanting a balance between AI and human contact. Some noted that AI is built by people, so human flaws can seep into the technology, and one drew a line between trusting the technology and trusting its programmers.
Trust in Online Technology Shows No Clear Split by Mental Health Status
Worries about privacy, data use, and the power of big companies showed up across participants. The authors suggest distrust may stem more from broad experiences with technology than from mental health status. That conclusion has limits. Only two participants had psychosis, and the researchers say it remains plausible that the roots of mistrust differ between groups.
Participants mostly talked about the internet in general rather than formal health apps, so any link to clinical care is a possible extension, not a direct finding. Still, the authors argue that developers should build trustworthiness into design instead of assuming it will develop through use, and that clinicians recommending digital tools may need to discuss data security and privacy openly.
Popularity may be a poor stand-in for trust: a health app or website can attract millions of users without earning their confidence, and the people clicking may be the first to say so.
Disclaimer: This article summarizes one small interview study and is for general information only. It is not medical, mental health, or professional advice. Anyone with health or privacy concerns should consult a qualified professional.
Paper Notes
Limitations
Only 22 people were interviewed, and the sample was mostly white, with most of those reporting a mental health diagnosis having depression rather than psychosis. Recruitment for a wider range of diagnoses proved difficult, which limits conclusions about experiences across the full mental health spectrum. Diagnoses were self-reported rather than confirmed by a clinician, though the authors cite earlier research supporting self-report as a good indicator in online studies. Recruitment happened online and every participant chose an online interview, so the group likely leans toward people already comfortable with digital tools and leaves out those with limited digital access or skills. The brain-training app was a research tool rather than part of a care plan, and most participants could not see its immediate usefulness. Some may also have assumed it used AI, which could have shaped their views. Participants mostly discussed the internet broadly rather than specific health apps, so applying the results to clinical care is an extension. The authors also propose adding trust and emotional reactions to a widely used technology-adoption model, but those changes have not been empirically tested, and they say the results are not intended to be generalized to wider populations.
Funding and Disclosures
Funding came from the UKIR Trust in Autonomous Systems Hub Trust Node (grant EP/V026682/1), via the University of Southampton. The authors declared no conflicts of interest, and the funders had no role in the study’s design, data collection, analysis or interpretation, the writing of the manuscript, or the decision to publish the results.
Publication Details
Emma Claire Palmer-Cooper (corresponding author, University of Southampton), Emma Gentry, Elena Nichele, Tayyaba Azim, Mohammad Naiseh, Liz Dowthwaite, Mohamad Said Armouch, Frances Greenway, Mark Weal, and Ann-Marie Hughes authored the study, titled “‘I Don’t Trust It, but I Have to Trust It’: Trust vs. Use of Online Technology Across the Mental Health Spectrum in the UK.” It was published October 22, 2026, in the journal Behavioral Sciences (MDPI), 2026, Volume 16, Issue 11, Article 1969. Academic Editor: Hua Pang. DOI: https://doi.org/10.3390/bs16111969
