47% of Distressed Adults Turn to AI for Mental Health Support as Use Surges Among Vulnerable Groups

Adults with Severe Distress Increasingly Rely on AI for Therapy

A groundbreaking study from Tel-Hai University of Kiryat Shmona reveals that nearly one in three Israeli adults using generative AI platforms turn to tools like ChatGPT and Gemini for emotional support, coping advice, or mental health assistance1

1

. The research, published in Cyberpsychology, Behavior, and Social Networking, surveyed 584 adults aged 18 to 70 and found that 29.3% of generative AI users sought mental health support from these platforms at least once in the past year1

1

. Among individuals experiencing severe psychological distress, that proportion surged dramatically to 47%1

1

.

Led by Dr. Dorit Hadar Shoval from the School of Psychology, the investigation identified psychological distress as the strongest predictor of AI-mediated help-seeking, with younger adults and women significantly more likely to confide in AI chatbots for mental health advice1

1

. The study also examined attachment styles, finding that adults with elevated attachment anxiety initially showed higher inclination toward AI for emotional reassurance, though this ceased to be an independent predictor once baseline distress was factored into regression models1

1

.

Adolescents Turn to LLM Chatbots Despite Wanting More Help

Parallel research examining over 24,000 students in England through the 2025 OxWell Student Survey found that 2.7% of adolescents reported using general-purpose AI for mental health support in the past year2

2

. The survey, conducted between February and May 2025 among students with a mean age of 14.0 years, revealed that 55.9% of those who rated their experiences found AI for therapy helpful, while 31.4% found it neither helpful nor unhelpful, and 12.8% found it unhelpful2

2

.

Crucially, 63.2% of adolescent AI users indicated they would have liked additional mental health support during the previous year, pointing to significant unmet needs2

2

. Most adolescents using these platforms had accessed other support forms, with 92.1% reporting support elsewhere and 46.1% accessing formal services2

2

. However, 16.5% of AI users either relied solely on these platforms or combined them only with online resources, making them potentially invisible to conventional services2

2

.

Vulnerable Demographics Show Higher AI Adoption Rates

The adolescent study identified several characteristics associated with AI for emotional support use. Transgender and gender-diverse adolescents showed higher odds of platform use compared to boys and young men2

2

. AI chatbots for mental health were more common among those reporting frequent food insecurity, higher depression and anxiety symptoms, previous self-harm, frequent loneliness, or occasional bullying2

2

. Adolescents from Asian or Asian British and other minoritized ethnic groups demonstrated higher odds of use, with the strongest associations observed in these ethnic categories2

2

.

Asian or Asian British adolescents were significantly more likely to rate the support helpful and also more likely to report unmet support needs, alongside those with higher depression and anxiety scores and previous self-harm history2

2

.

Critical Gaps in Regulation and Safety Protocols Emerge

The rapid adoption of conversational agents as informal therapeutic proxies exposes a glaring regulatory void. Dr. Hadar Shoval emphasized the urgency: "These tools offer immediate, accessible, and nonjudgmental interactions, but they were not designed to function as mental health providers. The high level of use among distressed individuals highlights the urgent need to better understand both the possible benefits and the potential risks of relying on AI during periods of emotional vulnerability"1

1

.

While dedicated digital therapeutics and licensed teletherapy platforms operate under strict medical device standards, patient confidentiality rules, and emergency escalations for suicidal ideation, commercial consumer LLM chatbots carry no clinical mandate1

1

. These platforms, trained primarily on broad internet text to produce statistically plausible responses, lack psychiatric validation, standardized crisis protocols, and mechanisms to manage acute psychological crises or self-harm risks1

1

.

The findings raise immediate questions about what happens when vulnerable individuals experiencing severe psychological distress turn to unregulated AI systems during mental health crises. As millions discover that large language models offer an always-available, endlessly patient, and completely nonjudgmental conversational partner, the gap between user expectations and platform capabilities becomes a pressing concern for both developers and regulators. Watch for potential regulatory frameworks addressing AI for mental health as adoption continues accelerating among the most vulnerable populations.