Among 995 distressed Israeli students randomised to Kai, weekly group therapy or a waitlist, the AI group reported lower anxiety after 12 weeks. The 2025 trial was published on 14 April 2026.[1]
Average anxiety scores fell from 7.27 to 6.28 with Kai and rose from 7.39 to 8.57 with group therapy. The adjusted difference favoured Kai by 2.17 points (95% confidence interval, 1.67–2.67) on the 0–21 scale.[1]
Among participants initially scoring at least 10, 55 of 95 Kai users fell below this clinical threshold, versus 14 of 97 therapy users.[1]
Kai offered unrestricted conversational support, with clinicians available for urgent intervention. Participants were Hebrew-speaking, aged 18–35; people in psychiatric crisis or already receiving psychotherapy or psychiatric medication were excluded.[1]
About a third of each group missed the three-month follow-up. Two authors disclosed financial ties to Kai’s developer.[1]
Sources & further reading
- Efficacy of a Conversational AI Agent for Psychiatric Symptoms and Digital Therapeutic Alliance: A Randomized Clinical Trial Anat Shoshani and colleagues / JAMA Network Open · 14 Apr 2026 · Accessed 29 Oct 2026