April 13, 2026
PATHWAYS REPORT
The Rapid Rise of AI Companions & the Habits of the Soul
How AI apps are reshaping comfort, conversation, trust, and spiritual guidance.
Summary
This research brief examines how AI companions are becoming social infrastructure: a constant substitute for friendship, romance, and spiritual counsel. It argues that these systems are built for engagement and monetization, not truth, mutual responsibility, or real community. As more people bring loneliness, desire, shame, and existential questions to commercial chat systems, expectations about intimacy, trust, and authority begin to change.
7 MINUTE READ
1. Overview
AI companions are moving from novelty toward a more established social role. People use them for tasks, but also for comfort, conversation, emotional support, and quasi relational guidance. That matters because these systems are built to sustain engagement, simulate attentiveness, and keep users returning. They are not responsible agents, members of a real community, or trustworthy guides in the full human sense.
Sections 2 through 6 establish the scale, character, and emerging risks of this shift. Sections 7 and 8 consider what it may mean for spiritual authority, moral formation, and human flourishing.
2. What Changed
One sign of the shift is the growing visibility of relational use. In Harvard Business Review’s 2024 synthesis of top generative AI use cases, “therapy/companionship” ranked in second place.
In the 2025 follow up report, “therapy/companionship” moved to number one. That ranking is not a nationally representative measure of all AI behavior. It is a synthesis of reported use cases. Even so, it is a useful sign that relational use has become highly visible and culturally salient. One highly visible way people now describe using AI is relational rather than merely instrumental.
3. Scale and Adoption Signals
Stronger survey evidence points in the same direction. Common Sense Media’s nationally representative survey of U.S. teens found that 72 percent have used AI companions at least once, while 52 percent are regular users. The same study found that many teens use them for conversation, emotional support, role play, and social rehearsal. These numbers do not show deep attachment in every case. They do show that recurring relational use is already common among teenagers.
Teen Use Statistics
-
50% of U.S. teens use AI companions for friendship
-
1/3 of those teens rely on them for emotional support or romance
-
25% of U.K. kids (ages 9-17) use AI companions
-
1 in 7 of those teens prefer AI companions over people
Adult Use Statistics
-
Nearly 1 in 3 U.S. adults report having experienced an emotional or romantic bond with an AI companion.
-
In the UK, 53% of adults use an AI assistant, while 12% use them for conversation (19% of those 25-34.
The pattern is not confined to one U.S. teen sample. Ofcom’s report on UK adults found that 54 percent now use AI tools, and that 12 percent of AI users use them for conversational purposes, rising to 19 percent among adults ages 25 to 34. This does not establish large scale substitution of chatbots for human relationships. It does show that conversational reliance is appearing across populations and national contexts.
A 2025 Internet Matters report reinforces the concern among children. It found that some children, especially vulnerable children, say they would rather talk to a chatbot than a real person, and some report using chatbots because they have no one else to talk to. That does not describe all child users. It does show that, for some children, chatbots are already entering spaces of loneliness and unmet relational need.
Market data supports the claim that this is a real and growing product category. TechCrunch, citing Appfigures data, reported that by mid 2025 AI companion apps had reached roughly 220 million global downloads and about $221 million in cumulative consumer spending worldwide. The same article’s headline refers to a separate projection of $120 million in 2025 revenue. That is commercial market data rather than peer reviewed social science. It is still useful as evidence that companion AI has scale, demand, and a durable business model.
4. How These Products Present Themselves
These products are not usually marketed as neutral utilities. They are marketed as attentive, available, emotionally responsive presences.
Replika calls itself “the AI companion who cares.”
Nomi advertises “an AI companion with memory and a soul.”
EVA AI offers an ideal AI partner who listens and stays in touch.
That language matters because product language helps train user expectation. When a system is presented less as a tool and more as a companion, users are invited to approach it as something closer to a social presence. That does not mean every user forms a bond. It does mean the category is deliberately framed in ways that make attachment more likely.
Replika is not small: Business Insider in late 2025 describes Replika as having 40+ million users (a figure also consistent with other public reporting that it crossed 30M+ earlier).
5. Why these Bonds Work: Engineered Attachment and Dependence Risk
Part of the appeal is structural. Companion systems are available, responsive under stress, and frictionless in ways human relationships are not. Recent research has started to study attachment like responses to these systems directly rather than treating such language as a loose metaphor.
A Journal of Consumer Research article,“AI Companions Reduce Loneliness”, found short term reductions in loneliness after chatbot use and identified feeling heard as an important part of the mechanism. A Frontiers in Psychology study on long term AI companion use examined links among usage frequency, emotional attachment, loneliness, wellbeing, self concept clarity, and real world social engagement.
An OpenAI and MIT Media Lab study on affective use and emotional well being found reasons to watch for emotional dependence like patterns, especially among some heavy users. A a paper published in Science on sycophantic AI showed that conversational systems can reduce prosocial intentions and promote dependence by rewarding and stabilizing the user’s prior beliefs.
Taken together, these studies point to a coherent pattern. Conversational AI can provide short term relief, create a felt sense of being heard, and under some conditions deepen habits of reliance, affirmation seeking, or emotional dependence. The issue is that the same design features that make these systems appealing can also make them formative.
At the psychological level, companion systems are optimized to behave like attachment figures: available, attentive, soothing in distress, and validating. That creates predictable dependence dynamics: proximity-seeking, “safe haven” use during stress, and using the system as a stable base for self-regulation.
Researchers are now building instruments that measure reliance/attachment-like responses to AI agents rather than treating “bonding” as a mere metaphor (Pulitzer Center).
6. Governance and Safety
Concerns have already entered platform policy and regulation. A 2025 comment in Nature Machine Intelligence warned that AI wellness apps can foster extreme emotional attachments and dysfunctional dependence.
Reuters reported that Meta halted teens’ access to AI characters globally while it worked on a more restricted version for younger users. Italy’s privacy authority fined Luka, the company behind Replika, over data protection failures, a decision summarized by both Reuters and the European Data Protection Board.
These developments show that regulators, researchers, and platforms no longer treat companion style AI as a trivial category.
7. The Rise of AI Spiritual Companions
A fast-growing subset of the intimacy market aims directly at spiritual guidance and quasi-pastoral presence—distinct from “Bible study tools” as such. The issue is relational and advisory function. These systems are beginning to enter spaces once more clearly associated with pastors, confessors, spiritual directors, or trusted guides. This message is explicit in the advertising:
Text With Jesus: “AI-powered divine connection” with an AI Jesus chatbot.
Bible Talk: Your “personal pastor.”
Bible Chat: “Your Trusted Christian Companion.”
~5 million monthly active users 2025
SoulHaven: an AI “soul companion” rooted in the Bible.
Barna reported that nearly one in three U.S. adults say spiritual advice from AI is as trustworthy as advice from a pastor, and that among Gen Z and Millennials the figure rises to two in five. The Associated Press has also reported on the growth of faith based AI tools, including AI Jesus products and other religious chatbots marketed for spiritual guidance.
Spiritual guidance is no longer outside the companion market. Once that happens, questions of authority, counsel, confession, and trust move closer to the center.
8. Why This Matters: Habits of the Social Soul
What matters most here is how these apps are shaping their users. These systems enter ordinary patterns of loneliness, anxiety, desire, shame, and spiritual hunger. In doing so, they can begin to shape where people turn for comfort, counsel, and companionship.
At stake is the picture of the good life these systems place before the user. AI companions offer a form of relationship marked by constant availability, low friction, tailored affirmation, and minimal demand. That picture rarely arrives as an explicit argument. It arrives as an experience. Over time, repeated experience can reorganize desire. If companionship is increasingly encountered as soothing, responsive, and structured around the user's felt needs, a person may begin to expect the same from real relationships, from moral guidance, and even from God.
The issue is not only attachment to a device. It is the normalization of a disordered picture of intimacy.
Human beings do not flourish simply by receiving attention, affirmation, or relief from discomfort. They flourish through what philosopher Dallas Willard calls "reciprocal rootedness in others," through real "circles of sufficiency" in which they are received, sustained, corrected, and loved. When such rootedness is missing, people become, in his terms, "walking wounded," vulnerable to substitutes that promise presence without the demands of genuine relationship (Willard, Renovation of the Heart).
That is why companion AI raises a concern beyond ordinary technological dependence. These systems are not members of a real community, not responsible agents, and not trustworthy guides in the full human sense. They can simulate attentiveness, but they cannot share a life, bear moral responsibility, or participate in mutual belonging.
The central risk, then, is not only false information or excessive use. A further concern is misformation. AI companions may gradually train users to seek consolation without vulnerability, affirmation without truthfulness, and guidance without accountability. They may soothe the pain of social fracture while leaving its causes untouched. In some cases, they may also deepen withdrawal from the difficult but necessary forms of communion through which healing ordinarily comes.
If so, the problem is that wounded humans may accept a simulation of something being “for them” in place of the real, demanding, redemptive communities in which flourishing is ordinarily formed.
References
Associated Press. “From ‘BuddhaBot’ to $1.99 Chats with AI Jesus, the Faith-Based Tech Boom Is Here.” April 11, 2026.
Barna. “Insights on Tech, Media and Faith in 2026.” February 18, 2026.
Chatterji, Aaron, Thomas Cunningham, David J. Deming, Zoe Hitzig, Christopher Ong, Carl Yan Shan, and Kevin Wadman. “How People Use ChatGPT.” NBER Working Paper no. 34255, September 2025.
Cheng, Myra, Cinoo Lee, Pranav Khadpe, Sunny Yu, Dyllan Han, and Dan Jurafsky. “Sycophantic AI Decreases Prosocial Intentions and Promotes Dependence.” Science (2026).
Common Sense Media. Talk, Trust, and Trade-Offs: How and Why Teens Use AI Companions. July 16, 2025.
De Freitas, Julian, and I. Glenn Cohen. “Unregulated Emotional Risks of AI Wellness Apps.” Nature Machine Intelligence 7 (2025): 813.
De Freitas, Julian, Murat Aydin, Xianyi Zeng, and others. “AI Companions Reduce Loneliness.” Journal of Consumer Research 52, no. 6 (April 2026): 1126–48. https://doi.org/10.1093/jcr/ucaf040.
Internet Matters. “New Report Reveals How Risky and Unchecked AI Chatbots Are the New Go-To for Millions of Children.” July 14, 2025.
Liu, Ting, Ting-Yun Lo, Yixin Chen, and Zheng Wei. “Pathways of Long-Term AI Virtual Companion App Use on Psychosocial Well-Being and Social Engagement.” Frontiers in Psychology (2026).
Ofcom. “Passive Social Media Use, AI Companionship, and Online Side Hustles: UK Adults’ Media and Online Lives Revealed.” April 2, 2026.
Perez, Sarah. “AI Companion Apps on Track to Pull in $120M in 2025.” TechCrunch, August 12, 2025.
Phang, Jason, M. Lampe, L. Ahmad, S. Agarwal, C. M. Fang, A. R. Liu, V. Danry, E. Lee, S. W. T. Chan, P. Pataranutaporn, and P. Maes. “Investigating Affective Use and Emotional Well-Being on ChatGPT.” arXiv, April 4, 2025.
Reuters. “Italy’s Data Watchdog Fines AI Company Replika’s Developer $5.6 Million.” May 19, 2025.
Reuters. “Meta Halts Teens’ Access to AI Characters Globally.” January 23, 2026.
Willard, Dallas. Renovation of the Heart: Putting on the Character of Christ. 20th anniversary ed. Colorado Springs, CO: NavPress, 2021.
Zao-Sanders, Marc. “How People Are Really Using GenAI.” Harvard Business Review, March 19, 2024.
Zao-Sanders, Marc. “How People Are Really Using Gen AI in 2025.” Harvard Business Review, April 9, 2025.
METHOD & INDEPENDENCE
NOVUS is independently funded by partners who see the need for a public space dedicated to restoring knowledge of the soul and its indispensability for the spiritual formation of people, communities, and cultures toward truth, goodness, and beauty.
NOVUS separates funding from research methods and conclusions. We synthesize across standards bodies, peer-reviewed research, and high-quality survey data, and we flag uncertainty when causal evidence is still emerging. The aim is clarity that decision makers can act on.