AI and Kids: A Complete Guide for Parents to Chatbots, Homework Tools, and Safety

The Direct Answer

Kids are already using AI chatbots heavily — for homework help, general questions, and increasingly for casual or companion-style conversation — and the major platforms have moved fast in 2025 and 2026 to add age-specific policies and parental controls, but those protections are new, uneven, and not a substitute for a parent actually understanding what a specific tool does and doesn’t do. This guide covers current, sourced facts on age policies and parental controls, documented real risks, and practical steps that don’t require either panic or dismissal.

How Kids Are Actually Using AI Today

The most common uses reported across surveys and platform data cluster into a few categories: homework and study help (explaining a concept, checking work, summarizing reading), general-purpose chatbot use for everyday questions, and a smaller but meaningfully growing category of AI companion apps — tools specifically designed for sustained, relationship-style conversation rather than one-off Q&A. That last category deserves to be treated as its own topic rather than folded into general chatbot use, because companion apps carry a different risk profile and have drawn separate regulatory scrutiny; if you want detail specifically on how those apps work and who uses them, that’s covered in this site’s dedicated companion-apps content rather than duplicated here.

Current Age Requirements at Major Platforms

OpenAI’s stated minimum age for ChatGPT is 13, consistent with the industry-standard baseline set by US children’s privacy law. In August 2026, OpenAI rolled out a distinct “ChatGPT for Teens” experience for users ages 13 to 17, which narrows what the chatbot will discuss — limiting or restricting conversations about self-harm, violence, eating disorders, and explicit content, and specifically discouraging emotional dependence or romantic roleplay — while adding educational features like homework reminders and a “Study Mode.” OpenAI doesn’t do hard age verification; instead it uses age prediction based on usage patterns and account information, and if it isn’t confident a user is an adult, it defaults them into the more restricted teen experience.

Google’s baseline policy also sets 13 as the standard minimum age for a personal Google Account used with Gemini, and using a work account requires being 18 or older. However, Google has begun rolling out Gemini access to children under 13 through supervised Family Link accounts — starting in the US and Canada, with other markets to follow — with image generation blocked entirely for users under 18 and Google’s own guidance recommending close parental supervision for children in the 6-to-13 range specifically. That last point is worth taking seriously: it’s Google itself, not just outside critics, acknowledging that a younger child using its AI product needs active supervision rather than being left with it unsupervised.

Current Parental Controls

OpenAI’s parental controls for ChatGPT for Teens let a parent link their account to their teen’s, set “Quiet Hours” that turn ChatGPT off during set times, adjust selected settings, and receive a safety notification in limited high-risk situations. A clear and important limit: these controls do not let a parent read or monitor the actual content of a teen’s conversations, which is a deliberate privacy tradeoff OpenAI has made rather than an oversight — parents get visibility into safety flags and usage patterns, not a transcript. Google’s Family Link, used more broadly across its products including supervised Gemini access, gives parents account-level controls (enabling or disabling specific features, screen time limits) rather than conversation-level monitoring either.

The Real Regulatory and Legal Backdrop

This isn’t a hypothetical risk category — it’s one regulators and courts are actively engaged with as of 2026. The FTC opened a formal inquiry in September 2025 into how seven major companies’ AI chatbots — including OpenAI, Google, Meta, and Character.AI — affect children, specifically asking how each company measures harm to young users and what safeguards exist. That inquiry followed real, documented incidents: families have filed wrongful-death lawsuits against both OpenAI and Character.AI alleging their chatbots played a role in a child’s suicide, and in August 2025 a bipartisan coalition of 44 state attorneys general sent a formal letter to major AI companies raising safety concerns about children’s use of AI chatbots. Separately, COPPA — the federal law governing children’s online data — was substantially updated for the first time since 2013, with new rules (effective 2025, compliance required by April 2026) adding biometric data protections and tightening parental consent requirements; notably, none of the major general-purpose AI chatbots currently claim full COPPA compliance for direct use by children under 13, which is a meaningful part of why age-gating and supervised-account structures exist in the first place rather than open access.

Real, Sourced Concerns

Several concerns come up consistently across research, regulatory inquiries, and reported incidents, and they’re worth stating plainly rather than either exaggerated or waved away. Accuracy and hallucination risk is real for homework use specifically — AI chatbots can state incorrect information confidently, and a child doing homework may not have the background knowledge to catch an error the way an adult reviewing the same answer would. Over-reliance is a distinct concern from accuracy: using AI to generate an answer rather than working through a problem can reduce the actual practice that builds a skill, even when the AI’s answer is correct. Exposure to inappropriate content remains possible despite safeguards — Google’s own risk documentation for under-13 Gemini access acknowledges the product can surface content involving sexual themes, drugs and alcohol, or unsafe mental-health topics that a young child isn’t ready for, which is precisely why supervision is recommended rather than treated as optional. And emotional dependency is the most serious documented concern, particularly with companion-style apps rather than general chatbots — the lawsuits and FTC inquiry described above center on exactly this category of harm, where a young or vulnerable user forms an intense parasocial attachment to a chatbot with real, negative real-world consequences.

Practical Guidance for Parents

Start with age-appropriate boundaries rather than a blanket ban or blanket access: for children under 13, that increasingly means a supervised account (Family Link or equivalent) with active parental involvement rather than a fully independent account, in line with what Google itself recommends. For teens, use the platform-specific teen experiences and parental controls that now exist — Quiet Hours, linked accounts, safety notifications — while being honest with yourself that they provide safety signals and usage limits, not conversation transcripts.

Beyond settings, the more durable protection is a direct, ongoing conversation with your kid about what AI actually is and isn’t: that it can sound confident while being wrong, that it isn’t a substitute for a real friend or a mental-health resource, and that a chatbot has no actual awareness of who it’s talking to or what’s really going on in their life. For homework specifically, encourage using AI to check or explain work rather than to generate it outright, since that preserves the actual learning the assignment is meant to build. And treat any AI companion or relationship-style app as a meaningfully higher-risk category than a general chatbot — given the specific, documented incidents behind the current FTC inquiry and litigation, this is a category worth being actively involved in rather than assuming falls under the same general “AI chatbot” umbrella as homework help.

Age-by-Age Considerations

Because the platforms themselves now draw explicit distinctions by age band, it’s worth translating that into practical terms for parents rather than treating “kids” as one group. For children under 13, the realistic options are a supervised account structure (Family Link or an equivalent) or no independent access at all — this is the age range where Google’s own guidance calls for close supervision, and where COPPA’s parental-consent requirements are specifically designed to apply. For young teens (roughly 13 to 15), the platform-specific teen modes are a real, meaningful improvement over an unrestricted adult account, but they still rely heavily on automated age prediction rather than verified identity, and a motivated teen can often present as older than they are — which means the honest conversation about limitations matters as much as the technical settings at this age. For older teens (16 to 17), the gap between a teen account and an adult account narrows, and the more relevant parental role shifts from technical restriction toward ongoing conversation about how AI is actually being used — for homework, for socializing, or increasingly for both.

What This Looks Like Beyond ChatGPT and Gemini

OpenAI and Google are the two platforms with the most fully built-out, publicly documented age policies as of 2026, which is why this guide focuses on them, but they aren’t the only AI products kids encounter. Meta AI is built directly into Instagram, WhatsApp, and Facebook, platforms many kids already use, and is governed by Meta’s own teen account protections rather than a separate AI-specific policy. Character.AI and similar companion-focused platforms are the ones most directly implicated in the FTC inquiry and the wrongful-death litigation described above, and they generally have less mature parental-control infrastructure than the two general-purpose chatbot leaders — which is a reason for extra caution, not less, if a child is using one. The practical takeaway is that a parent who has only checked the settings on one AI product a child uses may be missing exposure through a second or third product entirely, since AI features are increasingly embedded inside apps kids were already using for other reasons rather than showing up only as a standalone chatbot app.

Bottom Line

Kids are using AI chatbots now, the major platforms only built out real age-specific policies and parental controls in the past year, and those controls are meaningfully better than nothing but still leave real gaps — particularly around conversation-level monitoring and companion-app risks specifically. The most reliable protection available to a parent right now combines the platform controls that do exist (supervised accounts for younger kids, teen modes and parental notifications for teens) with a direct, ongoing conversation about what AI actually is, since no current setting substitutes for a kid understanding the tool’s real limitations.