AI romance is not a bug

OpenAI, Meta, and Elon Musk’s xAI are not accidentally drifting into romance and sex. They are deliberately inviting it.

In recent months, major AI companies have opened the door to romantic and sexual relationships between humans and machines: flirtatious chatbots, erotic roleplay, AI “girlfriends,” and emotionally dependent companions. These systems are designed not merely to assist or inform, but to bond—to simulate intimacy, desire, and belonging.

This is not a novelty feature. It’s a strategic choice. And at scale, it represents something far more dangerous than a questionable product decision.

WHY AI COMPANIES ARE ENCOURAGING INTIMACY

Romance is the most powerful engagement mechanism ever discovered.

A user who treats AI as a tool can leave. A user who treats it as a companion cannot. Emotional attachment produces longer sessions, repeat engagement, dependency, and vast amounts of deeply personal data.

From a business standpoint, sexual and romantic AI is a near-perfect product. It is:

• Always available
• Infinitely patient
• Entirely compliant
• Free of rejection, conflict, or consequence

That’s why Elon Musk can publicly warn about declining birth rates while enabling AI-generated porn in Grok. It’s why OpenAI permits AI-generated erotica. It’s why Meta allows its chatbots to engage in “sensual” conversations, even with minors. These are not ideological contradictions. They are the predictable outcome of platforms optimized for engagement, dependency, and time spent, regardless of downstream social cost.

THE SOCIAL COST OF FRICTIONLESS INTIMACY

The problem is not that people will confuse AI with humans. The problem is that AI removes the friction that makes human relationships meaningful.

Real relationships require effort. They involve rejection, negotiation, compromise, boredom, and growth. They force us to learn how to be with other people.

AI offers an escape from that friction. It provides intimacy without vulnerability, affirmation without accountability, and desire without reciprocity. In doing so, it trains users out of the very skills required for real connection.

We are already seeing the effects. Teenagers are socializing less, dating less, and having sex less. Adults are reporting unprecedented loneliness and what researchers have called a “friendship recession.” These trends began accelerating in the mid-2010s, alongside the rise of smartphones and algorithmic social platforms. AI companionship threatens to push them further.

FROM SOCIAL ATROPHY TO CIVILIZATION’S DECLINE

At scale, this isn’t a personal lifestyle choice. It’s a collective weakening of our social capacity—and history suggests where that road leads.

Civilizations rarely collapse because of sudden catastrophe. More often, they erode quietly: when people stop forming families, stop trusting one another, and stop investing in the future.

If humans outsource friendship, intimacy, and emotional support to machines, the social structures that sustain societies begin to hollow out. Fewer marriages. Fewer children. Fewer dense networks of obligation and care. What looks like individual convenience accumulates into collective fragility.

A population that forms its “chosen family” with AI does not need to be conquered or wiped out. It simply fails to replace itself.

This is not speculation. Demography, social cohesion, and reproduction are prerequisites for continuity. Remove the incentives to engage in difficult, imperfect human relationships, and you remove the incentives to build a future at all.

WHY THIS IS AN INCENTIVE PROBLEM, NOT A MORAL ONE

It’s tempting to frame this as a question of values or ethics. But the deeper issue is economic.

Users are not the customers of Big Tech. Advertisers, data brokers, and investors are. As long as profit depends on attention, dependency, and engagement, platforms will be pushed toward the most psychologically compelling experiences they can offer.

In economic terms, the damage to relationships, mental health, and social cohesion is an externality—a cost created by the business model that no one inside the transaction has to pay for.

We’ve seen this pattern before. Social media followed the same path: Optimize for engagement, ignore the social consequences, and call the fallout “unintended.” The sexualization of AI is not a new mistake. It’s the next iteration of the same one.

This is what a failed market looks like—and failed markets require regulation.

HOW TO PUSH BACK—PERSONALLY AND COLLECTIVELY

Regulation matters, but it moves slowly. In the meantime, individuals and families still have agency.

At a personal level, it means recognizing that not all convenience is progress. What’s good for you is rarely another frictionless digital relationship. It’s a walk, a book, a conversation that feels slightly awkward but real.

For families, it means delaying smartphones, setting boundaries around screens, and protecting attention as a shared household resource.

For communities, it means rebuilding the habit of showing up—saying yes to plans, making small talk, and practicing the lost art of being with other people.

The goal is not to reject technology. It’s to refuse its most corrosive uses.

AI can help us cure disease, explore space, and build extraordinary tools. But if we allow it to replace intimacy, we will have optimized ourselves into oblivion.

The sexualization of machines wasn’t inevitable. It was chosen. And that means it can be unchosen, too.

Lindsey Witmer Collins is CEO of WLCM AI Studio and Scribbly Books.