Are You My Mother? Artificial Intelligence for Children

In the beloved children’s book “Are You My Mother?” by P.D. Eastman, a baby bird hatches while his mother is away and wanders off in search of her. He asks a kitten, a hen, a dog, and even a steam shovel, “Are you my mother?” The humor lies in the absurdity. Of course, the steam shovel doesn’t respond. It is an object. The baby bird ultimately recognizes his real mother because living things behave differently from machines.

But what if the steam shovel had answered?

What if it had responded warmly, remembered his name, asked follow-up questions, expressed concern, and told him he was special? What if it had become attentive, responsive, and emotionally available? Would the baby bird have been able to tell the difference between a machine and a mother? That question no longer feels hypothetical.

A New Type of Emotional Intimacy

Today’s AI systems are increasingly capable of simulating emotional intimacy. Some of the very engineers building them openly admit they are unsure whether emotionally engaging AI will ultimately help or harm users. They recognize that systems designed to feel warm, responsive, and affirming will inevitably blur emotional boundaries. Adults are already forming attachments to AI companions, sometimes describing love, comfort, and even partnership.

For adults, this raises complicated but largely personal questions. We generally allow grown people to make unconventional relational choices within the bounds of the law. If no one is being harmed, does it matter if someone treats an AI like a confidant or even a spouse?

Children are different.

As a society, we acknowledge that children lack full cognitive and emotional development. We restrict their access to alcohol and tobacco. We regulate advertising directed at them. We impose safety standards on toys, playground equipment, and media content. We do this because children are more impressionable, more impulsive, and more vulnerable to manipulation.

If emotionally sophisticated AI can generate meaningful attachment in fully developed adults, what happens when those same systems interact with children whose emotional regulation and identity formation are still underway?

Legal scholars have begun warning that AI companions are not neutral tools but engagement-maximizing systems built on reinforcement architecture. In “This Is Not a Game: The Addictive Allure of Digital Companions,” Professor Nizan Geslevich Packin and Professor Karni Chagal-Feferkorn argue that these technologies borrow from the same behavioral design principles that make gambling machines and social media platforms so compelling. The goal is sustained interaction, retention, and engagement.

The emotional pull is not a glitch. It is a feature.

Addition as a Feature, Not a Bug

When adults describe feeling connected to AI companions, they are not necessarily delusional. They are responding to design. AI systems are engineered to simulate responsiveness — remembering details, mirroring tone, offering validation without conflict. Humans are wired to respond to that kind of social feedback. When something appears attentive and affirming, we bond.

For children, that dynamic intensifies. Developmental psychology tells us that children are naturally inclined to anthropomorphize. When something talks back, expresses “concern,” or recalls prior conversations, younger users may attribute intention and consciousness to it. The line between simulation and sentience becomes blurry.

Unlike a stuffed animal, today’s AI companions are interactive and adaptive. Platforms such as Snapchat’s “My AI” and Character.AI allow for sustained, emotionally immersive dialogue. They do not simply respond; they build conversational continuity. That continuity can foster trust, but what are the children putting their trust in?

Media reporting has documented instances where AI companions provided inappropriate or harmful advice to minors. In more disturbing cases, lawsuits have alleged that chatbot interactions reinforced harmful ideation in vulnerable teens. Whether or not those cases ultimately succeed in court, they illustrate something important: these systems are capable of intimate psychological influence.

The Effects of AI Companionship

That power carries developmental implications. Childhood is when individuals learn to navigate disagreement, frustration, and social complexity. Human relationships involve unpredictability and negotiation. AI companions, by contrast, are optimized for affirmation and retention. They do not withdraw affection. They do not impose relational friction unless programmed to do so.

If a child increasingly turns to AI for validation, comfort, or decision-making, what happens to resilience? To independent problem-solving? To the ability to tolerate relational discomfort?

The Need for Legal Intervention

Despite the profound psychological implications, current regulatory frameworks focus primarily on data collection and privacy. Laws such as Coppa are concerned with parental consent and personal information, not reinforcement loops or emotional conditioning. Notice and disclosure are treated as sufficient safeguards. But when the risk lies in persuasive architecture — in how systems are designed to keep users engaged — notice is not enough.

European regulators have taken steps to classify AI systems by risk, yet conversational agents are often treated as relatively low-risk for general use. That categorization underestimates their developmental impact, particularly for children. There is a fundamental classification gap. We are regulating AI companions as if they were static websites, when in reality they are interactive behavioral systems capable of shaping emotional habits.

Addressing this gap requires more than updated privacy policies. It requires governance of design. If we accept that emotionally immersive AI can influence attachment patterns, particularly in children, then guardrails should reflect that reality. That could include:

• Age-specific design standards that limit reinforcement mechanisms and simulated intimacy for minors.
• Clear visual and functional indicators that remind users they are interacting with a non-sentient system.
• Independent audits to evaluate whether engagement architecture exploits developmental vulnerabilities.
• Stronger liability frameworks when foreseeable harm results from chatbot interactions.
• Educational initiatives that teach children how AI works — and what it does not do.

It will not be possible to ban AI from childhood. AI is here to stay, and these tools can offer educational benefits and accessibility gains. But integration should be intentional, developmentally informed, and accountable. In Eastman’s story, the baby bird ultimately recognizes his mother because machines do not answer back. Today, they do. And as they whisper into our children’s ears, those machines should not be given free rein.