The Moment

As of July 2025, AI companion apps have been downloaded 220 million times globally. Downloads grew 88% year-over-year. The market generated $221 million in consumer spending over seven months. 72% of American teenagers have used AI for companionship. These are not niche products. They are mainstream.

And yet the research is beginning to document something the market would prefer to ignore.


The Story

In March 2025, researchers from MIT Media Lab and OpenAI published the results of a four-week randomized controlled trial involving nearly 1,000 participants. Their headline finding: higher daily usage correlated with higher loneliness, greater emotional dependence, and lower real-world socialization.

The people most drawn to using AI for emotional support were the people who were already the most lonely. And the more they used it, the lonelier they became.

This is not a fringe finding. It comes from a joint study by two of the most credible institutions in AI research, published by the researchers who built the product being studied. The AI companion market is not addressing loneliness. It is deepening it.


What It Reveals

The market is responding to a real hunger. The U.S. Surgeon General declared loneliness a public health epidemic. The WHO attributed 871,000 annual deaths to the effects of loneliness. What the AI companion market has correctly identified is the scale of the need. What it has failed to understand is the nature of it.

The loneliness people are experiencing is not the absence of stimulation. It is the absence of being genuinely known — by someone who is also known by you, who shows up imperfectly, who has something at stake in the relationship. AI companions offer constant availability, zero friction, and unlimited validation. What they cannot offer is reciprocity. They have nothing at stake. They cannot be wounded. They do not need you back.

The friction people are trying to avoid is exactly where intimacy lives. When someone struggles to understand you but keeps trying anyway — that is love. The AI companion removes the struggle. It also removes the love.


The Frame

Every major civilization in history has understood something the AI companion market is now disproving by experiment: human beings are built for genuine relationship, and substitutes — however sophisticated — do not satisfy the need they simulate.

The founding generation of this country understood that self-governance required a certain kind of citizen — formed by family, community, and conviction — capable of the hard, costly, interdependent relationships that hold a free society together. That formation has been degrading for decades. The AI companion market is not the cause of that degradation. It is the latest product category to profit from it.


What It Asks

The question the data raises is not whether AI companions are bad technology. It is what kind of people we are becoming — and what kind of communities are still producing people capable of the real thing.

The answer is not a policy or a regulation. It is formation. The communities producing people capable of genuine, costly, reciprocal relationship are the ones investing in the infrastructure that has always done that work: strong families, committed local communities, and the conviction that another person's flourishing is worth the inconvenience of your own.

That infrastructure still exists. It is producing people who do not need a frictionless simulation of love because they are living the real version. They are harder to find than an app. They are not harder to become.


The Guardian's Lens
A civilization that outsources its loneliness to an algorithm has already lost the plot. The hunger the AI companion market is monetizing is real — but the product cannot reach it, because what people are actually starving for is not more responsiveness. It is more accountability. The relationships that form character, build communities, and sustain a free people are the ones that cost something. A nation of people who have forgotten how to be genuinely known by one another is a nation that has forgotten how to govern itself.
Learn more at theguardianscross.org

Further Reading

  • MIT Technology Review — "AI Companions: 10 Breakthrough Technologies 2026" Read here →
  • OpenAI — Affective Use Study — Their own published findings. Read here →
  • TechCrunch — "AI Companion Apps on Track to Pull in $120M in 2025" Read here →

Share this post

Written by

Comments

Our Featured Articles