The research on AI companions and well-being is early, uneven, and easy to misrepresent. This post tries to summarize what the literature actually says — and what it doesn't — for adult users thinking about whether to use one.
If you are in crisis, contact 988 (US) or your local emergency services. Fostera is not a therapist, therapy app, or substitute for licensed mental-health care.
What the research actually says
Three findings recur across peer-reviewed studies and major outlets in 2024–2026.
AI companions can reduce reported loneliness in the short term. Multiple studies through 2024–2025 found measurable reductions in self-reported loneliness scores after weeks of regular AI companion use. The effect sizes vary but the direction is consistent.
The effect is strongest when AI use supplements human connection, not replaces it. Studies that controlled for time spent with humans found that users who substituted AI for human time showed worse outcomes than users who used AI alongside maintained human relationships. The framing matters: AI as supplement = good outcomes; AI as replacement = mixed-to-bad.
Long-term effects are not well-established. Most studies are weeks to months in duration. The category is too new for definitive long-term data. We don't yet know what regular AI companion use looks like at the five-year or ten-year mark.
A useful entry point: Brookings' 2024 analysis on AI companions and well-being, IEEE Spectrum coverage of empathy research, and reporting from the New York Times and The Atlantic on user experiences and platform safety.
Where AI companions help
Three use patterns recur across positive accounts in research and journalism.
Reflective companionship. Users who use AI as a thinking partner — to process the day, to work through decisions, to articulate things they're not ready to say to humans — report sustained benefit. The journaling-adjacent use case is consistently the strongest.
Bridging social skill development. For users who are working on social skills (introverts practicing conversation, autistic adults working on specific scripts, users in social-anxiety treatment), AI companions provide low-stakes practice that supports — not replaces — human work.
Continuity in transition periods. Users going through major life transitions (moves, divorces, career shifts, grief that doesn't meet criteria for clinical care) report value from AI companions as steady presences during instability. The continuity is the point: the AI is consistent when other things aren't.
Where AI companions don't help
Three patterns recur across mixed or negative accounts.
As a substitute for clinical care. Users in crisis, with diagnosed mental-health conditions, or in active suicidal ideation should be in clinical care. AI companions are not therapists. The research is unambiguous that AI is not a substitute for licensed mental-health care.
As a replacement for human relationships. Users who reduce time with humans to spend more time with AI show worse outcomes. The mechanism is straightforward: human relationships are weight-bearing in ways AI relationships are not, and substituting one for the other leaves the user lighter on social support, not heavier.
For minors. AI companions for users under 18 are a different product with different risks. The 2024–2025 lawsuits and the 2026 settlements made the case concrete. Fostera is 18+ adults-only and we do not serve minors.
What "use responsibly" means
The research suggests four practical guidelines.
Use AI companions as supplements, not replacements. Maintain human relationships actively. Treat AI as adding to your social and reflective life, not substituting for it.
Don't use AI as a substitute for clinical care. If you'd benefit from therapy, get therapy. AI companions can be useful adjuncts to therapeutic work — many users report using AI to journal between sessions — but they are not therapists.
Be honest with yourself about what you're using it for. Reflective companionship is one use; emotional avoidance is another. The research suggests the first helps and the second doesn't. Self-awareness about which one you're doing matters.
Pick a platform with clear safety practices. Enforced age-gating, stable content policy, visible memory controls, and a real crisis disclaimer are non-negotiable in 2026. See Is your AI companion safe in 2026? for the full guide.
Why we are 18+ only
Because the lawsuits, the regulatory actions, and the consumer-research literature converge: AI companions for minors are a different product, with a different risk profile.
Fostera is built for adults 18+. Age verification is enforced at signup. We do not serve minors and have no plans to. Adult users get a clearer product, a clearer policy, and a clearer relationship with the platform.
For more on Fostera's safety approach, see /safe-ai-companion.
The boundaries Fostera holds
Fostera is not a therapist, therapy app, or substitute for licensed mental-health care. We do not diagnose, treat, or claim therapeutic outcomes. Users showing crisis signals see 988 (US) and local emergency services surfaced in product.
Fostera is not a replacement for human relationships. AI companions are supplements, not substitutes. We design with that boundary in mind and articulate it clearly in onboarding.
Fostera is not for minors. 18+ adults-only, enforced.
These boundaries aren't marketing positioning. They're the lessons from the 2024–2025 case law and the consumer research, made operational.
Frequently asked questions
Do AI companions help with loneliness? Some users report benefit. Research suggests the effect is real in the short term and works best when AI supplements (not replaces) human connection.
Can an AI companion replace therapy? No. AI companions are not therapists or substitutes for licensed mental-health care. If you'd benefit from therapy, get therapy. AI can be a useful adjunct — many users journal with AI between sessions — but not a replacement.
Are AI companions safe for adults? On platforms with enforced age-gating, stable content policy, and visible memory controls — yes. Fostera meets that bar. Always check a platform's safety practices before signing up.
What does the research say about long-term AI companion use? Long-term effects are not well-established. Most studies are weeks to months. The category is too new for definitive long-term data.
Is Fostera safe? Fostera is 18+ adults-only with enforced age-gating, stable published content policy, visible memory controls, and an in-product crisis disclaimer.
What's the right way to use AI companions? As supplements to human connection, not replacements. For reflection, journaling, social-skill practice, and continuity during transitions — not for crisis support or clinical care.
Where can I read the research? Start with Brookings, IEEE Spectrum, the New York Times, and The Atlantic. Academic literature is searchable on Google Scholar; key terms include "AI companion well-being," "social robots loneliness," and "human-AI interaction longitudinal."