Morning edition · Friday, 30 October 2026
Three in ten asked the chatbot how they felt; nearly half of those in distress
An Israeli survey of people who use AI tools finds that the worse someone feels, the likelier they are to bring it to the machine. It does not say whether that helps, and it says so.
Some questions are easier to type than to say. A study published this week in Cyberpsychology, Behavior, and Social Networking asked 584 Israeli adults, aged 18 to 70, who had used a general-purpose AI tool such as ChatGPT at least once in the past year, whether they had ever used it for emotional support, coping advice or help with their mental health1. In all, 29.3% said yes1. Among participants with severe psychological distress, the figure was 47%1.
The pattern is the finding. Of the things the researchers measured, distress was the one most strongly linked to turning to AI for support; younger people and women were also significantly more likely to do so1. The team, led by Dorit Hadar Shoval of Tel-Hai University with Elad Refoua of Bar-Ilan University, also looked at attachment style. People with more attachment anxiety, a persistent worry about being let down by those close to them, were more likely to seek support from AI, but once distress was taken into account that link was no longer significant1. The two travel together, and the study cannot say which comes first1.
“These tools offer immediate, accessible and nonjudgmental interactions, but they were not designed to function as mental health providers,” Hadar Shoval said1. Consumer chatbots, the authors note, carry no clinical mandate and have no established safeguards for psychological crises1.
Read the limits as carefully as the headline. Participants came from an online panel in November 2025, so the figures describe this sample, not Israel’s population1. “Use” means anything from a single exchange to a regular habit1. And the study does not judge whether the conversations helped or harmed, or whether anyone used AI instead of a therapist1. It is a map of who knocks on the door, not a review of what happens inside.
That is still useful. If you work in a surgery, a school or a helpline, it suggests a plain question worth adding to the conversation: have you talked to an AI about this, and what did it tell you? The researchers call understanding such digital help-seeking “an important part of modern mental health care”1. The dull machinery of care, intake forms and follow-up calls, now has one more box to tick.
If you are struggling yourself, a chatbot can be a start; a person who can call you back is the better finish.
What to watch
- The authors say the next questions are safety, the quality of crisis responses, and whether AI complements professional care or in some cases replaces it1.
A tool that is always awake will be asked the questions people save for three in the morning.
Sources
- Nearly half of severely distressed generative AI users turn to chatbots for emotional support, study finds — Medical Xpress (Tel-Hai University release on Hadar Shoval et al., Cyberpsychology, Behavior, and Social Networking, DOI 10.1177/21522715261494423), 29 Oct 2026.
Every figure in this piece traces to a source above or to the desk’s own arithmetic on those sources, marked as such. Quotations are short and attributed. Spot an error? Write to the desk; corrections are logged on the method page.
Commentary for information only. Not investment, legal, tax, medical or political advice. Not an offer or solicitation. Analysis may be incomplete or wrong. Charts show published public data, not a trading feed; binding numbers live with the institutions that publish them.