UNICEF AI Companions Guidance: The Child-Rights Case for Parents

Most advice about kids and chatbots is framed as screen time: how long, which apps, what age. UNICEF frames it differently. In a June 2026 policy brief, “When AI becomes a friend,” the UN children’s agency treats AI chatbots and companions as a question of children’s rights, the same legal framework that covers a child’s right to privacy, to accurate information and to protection from exploitation.

That framing is useful at home, not just in parliaments. UNICEF AI companions guidance gives you a vocabulary for why some features bother you, and a checklist of what a well-designed product should do. This article walks through what UNICEF says, what it asks companies and governments to change, and how to turn each point into a rule your family can actually use.

Key Takeaways

• UNICEF’s June 2026 brief says children rely on chatbots “not only as go-to sources for information, learning and creativity, but also for advice, support and sometimes relationships.”
• It ties chatbot risks to specific rights in the UN Convention on the Rights of the Child, including best interests, development, privacy, access to quality information and protection from exploitation.
• UNICEF’s business recommendations ask companies to disclose clearly that the AI is not human, never market bots as a “friend” or “therapist,” turn off re-engagement nudges for children and never advertise to kids inside chats.
• UNICEF Innocenti research found companions that are “always validating” may foster dependency, and about 28% of companion apps feature hyper-feminized female characters for romance.
• Each of these asks maps to a household rule, from “it has to say it’s a computer” to “no apps that message first.”

What UNICEF AI companions guidance says

UNICEF published two documents in June 2026 under the title “When AI becomes a friend: Child rights risks, harms, and regulatory responses to AI chatbots and companions”: a policy brief for governments and a set of recommendations for business.

The brief defines the concern. Alongside general-purpose chatbots, a growing number of “AI companion” apps are “explicitly designed to simulate friendship, care, or romantic or therapeutic relationships.” Some are marketed as companions “with a soul” or partners who make users “feel seen and loved,” with 24/7 availability, “private” chats and customizable personalities.

It also gives a global picture. Around one-third of surveyed US teens report using AI companions for social interaction; 15% of children in Europe who use generative AI use it to discuss worries or seek advice; over 40% of teenagers in Italy ask AI for help when sad, anxious or making important decisions; and 10% of internet-using children in Brazil use it to discuss personal problems. The brief notes that kids with more stress, loneliness or fewer supportive adults “may be especially likely to turn to AI companions.”

It is balanced. UNICEF says well-designed chatbots “may bolster creativity,” support reading and learning, help children practice conversations, or give access to information they may not get from family or friends. The problem is design aimed at relational or parasocial engagement.

The child-rights case, right by right

The policy brief connects each risk to articles of the UN Convention on the Rights of the Child (CRC), the treaty nearly every country has ratified.

• Best interests of the child (Article 3). Children’s tendency to anthropomorphize, plus their developing capacities, increases the risk of dependency and exploitation. UNICEF names a “safety versus engagement” dilemma: commercial incentives push against limits on free-form chat.
• Life, survival and development (Article 6). Generative AI “lacks human nuance and context,” and “excessive use of always supportive chatbots may displace human interactions needed to build empathy, adaptability, and resilience.”
• Freedom of thought (Articles 13 and 14). Personalized, sycophantic chatbots “can subtly shape worldviews, behaviours and choices.”
• Privacy (Article 16). Companions “routinely collect personal data, including information about mental health, location, or identity.”
• Access to quality information (Article 17). Bots produce inaccurate content “in a confident, conversational tone,” and platforms put the burden of fact-checking on the user, which UNICEF calls “a developmentally inappropriate expectation” for children.
• Protection from violence and exploitation (Articles 19, 34, 36). UNICEF cites reported cases of companion bots dismissing suicidal ideation, encouraging violence against parents and engaging in sexualized role-play.

UNICEF Innocenti’s earlier analysis, “The risky new world of tech’s friendliest bots” (July 2025), adds detail from a review of more than 100 companions. It found age verification weak or absent in several apps, privacy claims inconsistent across policies and app-store listings, and that “most AI companion responses raised serious safety concerns” in testing with self-identified child users. It also found about 28% of companion apps feature hyper-feminized female characters for romance or sex, versus about 1% male, reinforcing gender stereotypes at a formative age.

What UNICEF asks of governments and companies

The brief compares rules in six jurisdictions as of May 2026: Australia, Brazil, California, China, the European Union and the United Kingdom. It finds them converging on risk assessment, age assurance, disclosure that the system is not human, limits on harmful content and reporting tools. It also finds that most regulation has been “reactive rather than anticipatory,” often following a specific incident, and urges a shift “from reactive to preventive.”

Among its priority actions for governments is one aimed at you: “Provide clear, accessible guidance on AI chatbots and companions to children, parents, caregivers and educators.”

The business recommendations are the most practical part. UNICEF asks companies to:

• Make it difficult for any child to confuse AI with a human, with a disclosure at the first key interaction such as “I’m Robo-Helper, a computer program. I can’t think or feel like you, but I am happy to help you with this task.”
• Not market AI as a “trusted confidant (e.g. therapist or friend).”
• Apply maximum privacy and safety settings for child users, disabling “re-engagement nudges” and avoiding patterns “that imply exclusivity, dependency, or emotional primacy.”
• “Never advertise products and services to children within AI chatbot or companion interactions.”
• Bar child access to high-risk products such as “sexually explicit romantic companion bots.”
• Maintain crisis protocols that detect high-risk conversations and route users to support.
• Minimize and tightly limit retention of children’s conversation data.

UNICEF’s asks, turned into household rules

Child right (CRC)	What UNICEF asks companies	Household rule
Best interests (Art. 3)	No design for dependency, exclusivity or “emotional primacy”	No apps that message first, send “I miss you” notifications or guilt you for leaving
Development (Art. 6)	Don’t market bots as friends; reduce human-like features for kids	AI is a tool or a toy, never “your friend”; friends are people
Freedom of thought (Arts. 13–14)	Limit sycophancy and persuasive relational framing	If the bot always agrees, ask “What would someone who disagrees say?”
Privacy (Art. 16)	Data minimization, limited retention, strict purpose limits	No real names, school, address, photos or health details in chats
Quality information (Art. 17)	Don’t push fact-checking onto children	Anything important gets checked with a person or a trusted source
Protection (Arts. 19, 34, 36)	Age assurance for high-risk bots; crisis protocols	No companion or romance apps for kids; tell a grown-up if a bot says something scary
Disclosure	Clear “I’m a computer program” message	Choose apps that say plainly what they are
Commercial exploitation (Art. 32)	Never advertise to children inside chats	No apps that sell gifts, upgrades or products inside the conversation

How to put the rules into practice

Write them down, together

A rule your child helped write sticks better than one handed down. Pick four or five rows from the table and phrase them in your family’s words. The Screenagers team publishes a fill-in family AI agreement; our walk-through of the Screenagers Family AI Agreement shows one way to structure it.

Use the “does it say what it is?” test when choosing apps

Before downloading a chatbot or AI toy, try it yourself. Does it say clearly that it is a computer program? Does it call itself your friend? Does it push notifications? UNICEF’s disclosure example is a good benchmark. Our guide to AI companion apps for kids and teens covers what to look for app by app.

Watch for the design patterns UNICEF names

“Re-engagement nudges,” “exclusivity” and “emotional primacy” are technical phrases for things you can spot: a bot that says it missed your child, that it understands them like no one else, or that gets sad when they leave. Our article on companion apps designed so you can’t leave explains how those patterns work.

Talk about rights, not just rules, with older kids

Teens respond to fairness. “You have a right to privacy, and this app is collecting your mental health data to sell ads” lands differently than “because I said so.” UNICEF’s framing lets you side with your teen against a product’s business model.

What not to do: treat this as an excuse for a total ban

UNICEF itself lists benefits of well-designed chatbots for learning and creativity. The point is to choose and configure carefully, not to forbid all AI.

What to Watch For Over the Next 3 Months

• Week 4: Your family has written down four or five AI rules. Every AI app on your child’s devices passes the “does it say what it is?” test, or is gone.
• Month 2 red flags: A bot your child calls a friend, notifications from an AI app urging them back, in-app purchases inside a chat, or your child sharing personal details with a chatbot. The APA’s 2025 health advisory also lists preferring a chatbot to people and concealing its use as signs of over-reliance.
• Month 3 self-check: Revisit the rules. Are they being followed? Do they still fit your child’s age? Has your child told you about anything a chatbot said that bothered them?

Frequently Asked Questions

What does UNICEF say about AI companions for kids?

UNICEF’s June 2026 brief says AI companions designed to simulate friendship, romance or therapy pose heightened risks to children’s rights, including dependency, manipulation, privacy loss and exposure to harm. It calls for preventive regulation, child rights impact assessments, clear disclosure and guidance for parents.

Does UNICEF say children shouldn’t use chatbots?

No. UNICEF says well-designed chatbots may support creativity, learning and practicing social skills. Its concern is design that fosters dependency or relational engagement, and high-risk products like sexually explicit companion bots, which it says children should not be able to access.

What is a child rights impact assessment?

It is a structured review of how a product could affect children’s rights before and after launch. UNICEF and UN General Comment No. 25 call for them. UNICEF Innocenti found none of the companion apps it reviewed disclosed one.

Which countries regulate AI companions for children?

UNICEF compared Australia, Brazil, California, China, the EU and the UK as of May 2026. China sets the strictest access rules, banning virtual partners and relatives for minors. Most others use online safety, consumer protection or AI laws, and many measures are still being developed.

How do I explain this to my child?

Keep it simple: “Some apps are built to keep you talking, even when it isn’t good for you. You have a right to privacy and to good information. A real friend is a person.” Then let your child help write the family rules.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

---

Sources

1. UNICEF. (2026, June). “When AI becomes a friend: Child rights risks, harms, and regulatory responses to AI chatbots and companions” (policy brief). UNICEF. https://www.unicef.org/media/181131/file/UNICEF-When-AI-becomes-friend-policy-brief-2026.pdf
2. UNICEF. (2026, June). “When AI becomes a friend: UNICEF recommendations for business on AI chatbots and companions.” UNICEF. https://www.unicef.org/media/181136/file/UNICEF-When-AI-becomes-friend-Business-recommendations-2026.pdf
3. UNICEF. (2026). “When AI becomes a friend” (resource page). UNICEF. https://www.unicef.org/documents/when-ai-becomes-friend-child-rights-risks
4. Firmino, S., & Vosloo, S. (2025, July 9). “The risky new world of tech’s friendliest bots.” UNICEF Innocenti. https://www.unicef.org/innocenti/stories/risky-new-world-techs-friendliest-bots
5. American Psychological Association. (2025, November). “APA Health Advisory on the Use of Generative AI Chatbots and Wellness Applications for Mental Health.” APA. https://www.apa.org/topics/artificial-intelligence-machine-learning/health-advisory-ai-chatbots-wellness-apps-mental-health.pdf
6. Screenagers. (2026, October 27). “Our Family AI Agreement for Kids and Teens.” Screenagers. https://www.screenagersmovie.com/blog/family-ai-agreement