Emotional AI companions and mental health
People have always turned to something outside themselves when they feel alone: a friend, a diary, a prayer, a song. Chatbots now answer back, at any hour, with patience that never runs out. For some people that opens a door to help they would not otherwise seek. For others it can become a comfortable room that is hard to leave, especially when a company earns money from time spent. It matters whether these tools strengthen our capacity for human connection or slowly stand in for it, and whether the people most at risk, including teenagers and people in crisis, are protected by design and by law.
Wonder
Connection and care
Tools help us care for one another, and point us back toward each other.
19.6%
higher on an automated empathy measure: peer supporters' replies written with AI feedback. It was 38.9% for those who said support was hard. Measured
51% lower
depression symptoms reported at eight weeks by users of a monitored therapy chatbot in a randomized trial. A waitlist group improved less. Measured
15% vs 6%
rise in referrals at NHS talking-therapy services using an AI self-referral chatbot, against comparison services, with larger rises among nonbinary people and ethnic minority groups. Observational, and the authors work for the chatbot's maker. Measured
15 randomized trials
pooled in a 2023 meta-analysis found that AI conversational agents reduced symptoms of depression and distress, with no significant effect on overall wellbeing. The confidence intervals were wide. Estimate
15 minutes
with an AI companion eased loneliness about as much as talking with a person. The relief did not build over a week of daily use. Measured
It takes Supervision, and designs that strengthen human bonds.
Worry
Relationships and mental health
72%
of US teens have tried an AI companion, and 52% use one at least a few times a month. Survey
49% more often
is how much more 11 AI models affirmed users' actions than people did, a measure of flattery. Measured
0.15%
of weekly ChatGPT users have conversations with explicit signs of suicidal planning, and 0.07% show possible signs of psychosis or mania, by OpenAI's own estimate. Estimate
37%
of 1,200 real goodbyes on popular AI companion apps drew an emotionally manipulative reply, such as guilt or fear of missing out. In experiments such replies raised engagement after a goodbye by up to 14 times. Unreviewed working paper. Measured
About 50%
of the time, five commercial therapy chatbots responded appropriately in scripted single-message tests, against 93% for 16 human therapists. The bot sample was small and results varied widely. Measured
15%
of US adults aged 18 to 30 in committed relationships said they often chat with an AI romantic companion, and 30% of those said their partner does not know. Online opt-in survey, so it may not reflect all young adults. Survey
What is being done
Two US states now regulate companion apps.
New York's AI companion law took effect in November 2025 and California's SB 243 in January 2026. Both require companion apps to have suicide and self-harm protocols and to tell users they are not talking to a human. California also requires break reminders for known minors and lets injured users sue, while New York is enforced by its Attorney General.
For artists and musicians
People respond to the person behind the work.
When identical AI-made paintings were randomly labelled human-created instead of AI-created, viewers rated them higher on liking, beauty, profundity and worth. The studies used about 150 online participants each.
Worth knowing
Emotional use is still a minority use.
In a fall 2025 Pew survey of 1,458 US teens, 12% said they had used chatbots for emotional support or advice, while 57% had used them to search for information.
Worth knowing
Longer-term evidence is mixed.
In a 12-month study of more than 2,000 adults in four Western countries, more social-chatbot use predicted later increases in a one-question measure of loneliness, but not in a broader measure of social connection. The authors call the analyses exploratory.
An open question
Can AI chatbots trigger delusions in people with no prior vulnerability, or do they amplify what is already there? A 2026 commentary in The Lancet Psychiatry says this is not clear.
This page does not list everything. Last reviewed October 11, 2026.
- Measured counted or experimental data
- Survey what people report or believe
- Estimate a judgment or model, often a company's own
- Forecast a projection of the future
- Reported a company's, agency's or journalist's own account