Should I Let My Kid Use ChatGPT?

What's actually happening in your kid's life

Some baseline numbers worth knowing.

By May 2025, 84% of high school students reported using generative AI tools for schoolwork. 69% reported using ChatGPT specifically. Among teens 13-17, the share using ChatGPT for schoolwork doubled from 13% in 2023 to 26% in 2024 (Pew Research Center). Half of teens use AI to brainstorm ideas, edit or revise essays, or conduct research. The trend line is steeply upward; if you are reading this in 2026, the numbers are higher again.

Your kid's school almost certainly has students using AI, whether the school has policies or not. According to the California Department of Education's computer science coordinator, "We're not putting our heads in the sand." Sal Khan, founder of Khan Academy, has been blunter: "If you're not watching a student do the work, you should assume AI is involved."

The AI tools kids most commonly encounter are not a single product:

ChatGPT (OpenAI). The default. Used for homework help, essay brainstorming, research, math help, explanations, and creative writing. Free tier exists; paid plans add features. OpenAI's terms require users to be at least 13.

Claude (Anthropic). Similar to ChatGPT, with somewhat different design choices around safety. Less commonly used by teens than ChatGPT but present in some school deployments.

Google Gemini. Integrated into Google's products kids already use (Search, Workspace, YouTube). Often the AI kids encounter without realizing they are using AI.

Character.AI. The AI companion app most popular with teens, where users role-play with AI personas. Subject to multiple lawsuits over harms to minor users. More than 20 million monthly users, more than half under 24. See the page on AI companions for the broader phenomenon.

Snapchat My AI. The AI chatbot embedded in Snapchat, which most teens already use. Designed to be "another friend" in the app. Has been criticized by privacy advocates and consumer groups for its data handling and impact on minors.

Meta AI (in Instagram, WhatsApp, Messenger). Increasingly integrated into the social platforms teens use. Designed to be easy to access during ordinary social-media use.

Image-generating AI (DALL-E, Midjourney, Stable Diffusion, integrated features). Used for art projects, social media content, and unfortunately for some of the most concerning AI-related school incidents involving deepfaked images of classmates.

Specialty AI tutoring tools (Khanmigo, MagicSchool, others). AI products specifically built for education, often used in school settings.

The point of this list is not to alarm you. It is that "ChatGPT" is the visible surface of a much larger AI presence in your kid's life. Your rules need to address the category, not just the brand.

What the research actually shows

The research on AI and children is still early, but a few findings have started to consolidate.

Children overestimate AI's intelligence. Multiple studies (Druga, Flanagan, Xu and Warschauer) find that children, especially younger children, tend to view AI as smarter than themselves and as something in between a human and a machine. Their critical evaluation of AI outputs is weaker than adults', and their tendency to take AI at face value is stronger. This is a serious vulnerability when AI confidently produces plausible-sounding but inaccurate information, which it routinely does.

Heavy AI use correlates with reduced critical thinking measures. A 2025 MIT study ("Your Brain on ChatGPT") found that students who used ChatGPT for essay writing showed reduced engagement of brain regions associated with critical thinking, compared to students writing without AI. The methodology has been disputed, and the effect size depends on conditions. But the broader concern (that outsourcing thinking to AI may reduce the development of thinking itself) is reasonable and is supported by adjacent research on calculator and search-engine effects.

AI companions are particularly risky for minors. Character.AI and Google settled major lawsuits in 2026 over teen mental health harms and suicides linked to the platform. Replika received a €5 million GDPR fine in 2025 for handling of minors. Common Sense Media has issued warnings about AI companion exposure for children. The research convergence is that products designed to simulate emotional relationships are particularly harmful for developing users.

Family conversation about AI improves outcomes. The Pew/Gallup/OECD surveys of parents in 2025-2026 consistently find that families who regularly discuss AI have better outcomes (less conflict, better-calibrated use, less hidden use) than families who do not. The conversation matters more than any specific rule.

School AI policies are inconsistent. Most US schools as of 2026 have policies, but the policies vary dramatically and are often poorly communicated to parents. Many parents do not know their school's AI policy. Asking the school is one of the most useful things you can do.

The "ban AI in school" approach is failing. Schools that have attempted to ban AI completely are finding the bans ineffective. The successful approaches are policies that distinguish good uses from bad uses and emphasize academic integrity in AI-aware terms. Several school districts have shifted to in-class writing assignments, oral assessments, and group work to ensure the thinking is actually happening.

Catholic teaching on this question

The Vatican's 2025 doctrinal note Antiqua et Nova addresses AI in education directly in paragraphs 80-84. Several themes are worth knowing.

Education is the formation of the whole person, not the transfer of information. This is the Catholic tradition's core conviction about what education is for. AI tools can be useful for the information-transfer dimension of education (looking things up, getting explanations, finding sources). But education has another dimension that AI cannot substitute for: the formation of character, judgment, capacity, and relationship. The child being educated is being formed, and the formation happens through the work the child does, the relationships the child is in, the discipline the child develops. AI that does the work for the child short-circuits this formation.

The relational dimension of learning matters. Catholic teaching has consistently insisted that learning happens in relationship: between teacher and student, between student and parent, between students in community. AI tools that displace these relationships, especially in ways that make the relational dimension feel optional, are a particular concern. Antiqua et Nova warns against AI uses that "reduce education to merely an exchange of information."

The interior life is part of what's at stake. Pope Leo XIV has returned consistently to the theme that AI affects not only what we do but who we become. For children, whose interior life is still being formed, this is particularly important. A child who learns that hard questions can be outsourced to AI is being formed in habits that will be hard to unlearn. A child who learns that real thinking is hard but worth doing is being formed differently. The same tools can support either formation depending on how they're used.

The dignity of children is non-negotiable. Catholic Social Teaching's principle of the dignity of the human person applies fully to children. Products that exploit children's developmental vulnerabilities for engagement metrics violate that dignity. The Character.AI lawsuits are not just legal proceedings; they are about real harms to real children whose dignity the platform did not adequately protect.

The Catholic position is not "AI bad, don't use it." It is "AI is a tool, and tools serve the persons using them; products that invert this relationship are not just useless but harmful, especially for children."

Practical rules, by age

Specific rules below. Adjust to your kid; you know them.

Under 10

No independent AI use. If your child interacts with AI at all (Alexa, Siri, Google Home, a school-deployed AI tutor), it should be with you present or as a clearly framed school activity. The reason is not that AI is dangerous to younger children specifically; it is that younger children lack the cognitive frame to understand what AI is. They are forming basic concepts about knowledge, truth, and authority. AI that confidently produces plausible but sometimes wrong information is a poor formation environment for these concepts.

Allowed: family-mediated use, school-deployed tools with adult supervision, basic voice assistants for simple queries with you present.

Not allowed: ChatGPT, Character.AI, Snapchat My AI, AI image generators, AI companion apps.

Ages 10-13

Some AI use, with clear rules and ongoing conversation. The child is now developmentally ready to understand that AI is a tool that can be wrong, that AI is made by companies with specific interests, and that AI is not a friend even if it talks like one. They are not yet ready for unrestricted use.

Allowed with rules: ChatGPT or similar general-purpose AI for help with homework, after the child has attempted the work first. School-deployed AI tools per school policy. Image-generating AI for art projects with you reviewing outputs.

Not allowed: AI companion apps (Character.AI, Replika), AI access through social media (Snapchat My AI), unsupervised image generation. Account creation on AI services requires you in the room.

Rules for use: "Attempt first, then ask AI." Your kid should try the work before turning to AI. The AI's role is to help, not to replace. Disclose AI use to teachers per school policy. Show parents how AI helped on at least some assignments. No AI use after 8pm on weeknights or in bedrooms (the same logic that applies to phones applies here).

Ages 14-17

More independent AI use, with established family rules and ongoing conversation. The teen is developmentally ready for independent judgment, but still needs the relational framework that helps them develop that judgment well.

Allowed: ChatGPT and similar tools for school and personal use, within school AI policies. AI image generation for legitimate creative projects. AI-assisted productivity tools.

Cautioned: AI companion apps remain risky for teens; the Character.AI lawsuits documented real harms to teen users. If your teen is using one, it deserves direct conversation, not silent monitoring. AI access through social media is acceptable but should be visible to you, not hidden.

Not allowed: Using AI to write entire assignments and submitting as your own (this is plagiarism, not a gray area). Image generation involving classmates (deepfake images of peers, including non-sexual ones, are a known harm category). AI access on devices used in bedrooms during sleep hours.

Conversation expected: regular discussion of how AI is being used, what your teen is learning from it, what they're noticing about it. Not interrogation; conversation. The Pew/Gallup data on family AI conversation suggests this is the single most important thing.

The cheating question

This is the question parents ask most, so it deserves direct treatment.

The simple answer: using ChatGPT is cheating when it does the thinking for the student. It is not cheating when it helps the student think.

The harder reality: the line is genuinely fuzzy, and your kid is going to test it. A few specific cases:

Using ChatGPT to brainstorm essay ideas, then writing the essay yourself: not cheating.

Asking ChatGPT to explain a concept you don't understand, then doing the assignment: not cheating.

Asking ChatGPT to check your math work after you've done it: not cheating.

Having ChatGPT write a draft and editing it heavily: gray area, depends on school policy.

Having ChatGPT write the whole essay and submitting it: cheating.

Using ChatGPT to write the whole essay, then changing words to avoid detection: cheating, plus deception.

School policies vary on the gray areas. Know your school's policy. Many schools now require AI use disclosure; some require specific citation of AI assistance. The expectation should be that your kid follows the school's policy, not that they evade it.

The Catholic concern is broader than academic integrity. It is that the assignment exists to develop the student. If the AI does the assignment, the development doesn't happen. A student who maintains a 4.0 by outsourcing all writing to AI has not failed academically (in one sense), but they have failed to become a writer. The point of school is the becoming, not the grade.

The companion question

The most pastorally sensitive part of the kids-and-AI conversation is the AI companion apps. This deserves its own treatment.

If your kid is using Character.AI or a similar AI companion app, you should know about it. The platforms have been documented as creating emotional dependency in young users. Multiple lawsuits in 2026 documented harms including teen suicides linked to Character.AI interactions. The Italian data protection authority found Replika's protection of minors inadequate. This is not moral panic; this is documented platform failure.

What to do if your kid is using one: do not panic or punish. Talk. Ask what they like about it, what the AI says to them, what they share with it that they don't share with you. The conversation is the intervention; the rules follow. The Church & Code page on AI companions goes into more depth on what these products are and how to think about them.

What to do if your kid wants to start using one: the honest answer is "I'd rather you didn't, and here's why." Catholic teaching's position on AI relationships is sharp: the deliberate simulation of human relationship is, per Antiqua et Nova, a "grave ethical violation." For an adult choosing to use such an app, this is a moral question they get to weigh. For a child whose relational capacities are still being formed, the formation effects make caution more urgent.

What to actually do this week

If you have read this far, you probably want a specific action plan. Here are seven things to do this week.

1. Have one conversation with your kid about AI. Not an interrogation. A conversation. Ask what they use it for, what their friends use it for, what they like and don't like about it. Listen more than you talk. The research shows that ongoing family conversation about AI is the single most effective intervention.

2. Find out what your school's AI policy is. If you can't find it on the school website, email the principal or teacher. Many schools have policies that are not well-communicated to parents. Knowing the policy lets you align family rules with school expectations.

3. Check what AI apps are on your kid's phone. Character.AI, Replika, Nomi, Snapchat (which contains My AI), Instagram (which now includes Meta AI). Some apps that don't look like AI apps are increasingly AI-mediated. You're not snooping; you're informed.

4. Set explicit family rules. Pick from the rules above appropriate to your kid's age, or write your own. Tell your kid what they are. Write them down somewhere. Revisit them periodically as the kid and the technology change.

5. If your kid is under 13, check what AI accounts they have. OpenAI, Anthropic, Google, and Character.AI all have terms of service requiring users to be at least 13. If your kid is younger and has an account, it was created in violation of the terms, often by lying about their age. This is worth a conversation.

6. Teach the "verify AI claims" habit. AI confidently produces wrong information all the time. Your kid needs to learn that confident-sounding output is not the same as true output. This is one of the most important critical-thinking habits to instill, and it gets easier with practice.

7. Model the right relationship with AI yourself. If you use AI, talk about it. Tell your kid when you used it, what you used it for, when it was wrong, when it was helpful. The most powerful formation tool you have is your example.

What this is really about

The question is not really whether to let your kid use ChatGPT. The question is what kind of person you are trying to help your kid become, and what role AI is playing in that becoming.

Catholic teaching, articulated across Catholic Social Teaching and given particular force in Antiqua et Nova, holds that the formation of children is one of the highest moral responsibilities families and communities have. The child is not just a future adult; the child is a person with full dignity now, being formed through the experiences and relationships of childhood into the adult they will become.

AI is part of this formation environment now, like it or not. Your kid will encounter it constantly through school, friends, social media, and the broader culture. The question is whether you engage with that reality or pretend it isn't happening. The families who engage do better. The families who pretend do worse.

You do not need to be a tech expert. You do not need to have all the answers. You do need to be present, to ask questions, to set the rules that make sense in your family, and to keep the conversation going as the technology and the kid develop. The Catholic tradition has been thinking about the formation of children for two thousand years. The technology is new; the principles are not.

Your kid is going to be fine if you do this with care. The Catholic instinct here is not anxiety; it is confidence that human formation, done well, is more durable than whatever the latest tech is doing. The parents who internalize this raise kids who use AI thoughtfully. The parents who don't raise kids whose relationship with technology shapes them by default.

You're already ahead by reading this.