I Analyzed Seven of the Largest AI Chatbots to Determine How Safe They Are for Children.
The safety of AI for both adults and children is coming under increasing scrutiny , with increasing questions and concerns raising about how beneficial or harmful these conversational bots are to our mental health—even without considering their energy consumption, ethical, or economic implications. At the same time, in a growing number of countries, apps and services are becoming age-restricted , as governments recognize that giving children access to AI apps, social media, and the internet in general may not be the best idea for their development and well-being.
Major AI-powered apps like ChatGPT and Gemini are at the forefront of these trends, and the companies behind them are taking steps they say will ensure children’s safety. If you want to know what children in your family may be exposed to and what you can do about it, here’s a quick overview.
ChatGPT
OpenAI’s ChatGPT has one of the most comprehensive parental control systems available. The minimum age for users is 13, and if you’re under 18, parental consent is required to open an account. Age is determined based on several signals , including the time of day you use ChatGPT and the topics of your conversations.
Accounts for people under 18 are automatically converted to ChatGPT teen accounts. The focus is on education and learning (with a dedicated learning mode) and “age-appropriate protections to reduce exposure to sensitive or potentially harmful content.” Built-in protection against “dangerous topics,” including sexual, romantic, or violent role-playing, graphic violence, and self-harm, complements the protections for all users, which limit chats devoted to physical violence and illegal activity.
These ChatGPT teen accounts can also be linked to a parent account. However, while parental consent is required for use by those under 18, ChatGPT clearly states that teens have the option to unlink them if they choose. If you manage to convince your child to keep your accounts linked, you can set time limits on the AI bot’s use and adjust some privacy settings. However, you won’t be able to view the conversations.
To set up parental controls in ChatGPT via a web browser , click your profile icon (bottom left), then select “Settings” > “Parental Controls” > “Add Family Member” and follow the instructions. It’s also worth remembering that ChatGPT is free to use, anonymously, in any web browser, although OpenAI states that users accessing its AI in this way are subject to the default under-18 restrictions (and cannot save chats or access the archive).
Twins
Gemini is integrated with all other Google features and follows the general rules for Google accounts: access by anyone under 13 is prohibited unless you are part of a family group and use an account managed by an adult. You can manage your family through the Family Link web portal , but Gemini-specific parental controls only allow you to enable or disable access to artificial intelligence (settings such as screen time apply to all enabled apps).
For all individuals under 18 (partially defined by an approximate age estimate), regardless of whether they are part of a managed family group, Google implements additional protections . These include preventing AI from assuming specific personalities, protecting against emotional dependence on AI, and preventing bullying and other forms of harassment. Additionally, measures are in place to restrict discussion of substances prohibited for those over 18, such as alcohol or e-cigarettes, and mental health issues, such as eating disorders.
These protections for minors are in addition to the standard restrictions in place for everyone on Gemini. These restrictions are designed to discourage discussions about self-harm, suicide, violence, and illegal activity, and to offer quick links to crisis support. However, Google itself acknowledges that these filters aren’t perfect .
As with ChatGPT, there’s a loophole of sorts: anyone can chat with Gemini for free in a web browser, without any verification. These conversations can’t be saved, and they quickly run into usage restrictions, but they’re protected only by general account restrictions, not by restrictions specific to minors.
Meta AI
Meta AI will appear in several apps likely used by young people and children, including WhatsApp and Instagram, and will also have its own standalone app on the web and on phones. Standard restrictions for Meta social media accounts also apply to its AI: a minimum age of 13 and restrictions for anyone under 18. Like many other platforms, it uses some AI capabilities to identify accounts used by minors.
If you are under 18 and have a Meta Teen account, you will be blocked if you attempt to engage Meta AI in sexual or romantic roleplay or discuss alcohol or drug abuse. In chats on topics such as eating disorders and self-harm, the AI will suggest you contact appropriate support services.
Meta offers comprehensive parental control settings for its apps, including Meta AI, but connecting a parent account for teens is optional. This is done through the Meta Family Center portal , and parents can receive reports on common topics their children discuss with the AI—though they don’t receive precise transcripts or prompts. Any discussion of self-harm or suicide may trigger an alert in the connected parent account.
All users, both under and over 18, are provided with support and emergency assistance for discussing suicide and self-harm. There are strict restrictions on discussing illegal and dangerous activities, including drug use and cyberbullying, as well as sexual and adult content.
AI Siri
Apple users must be over 13 years old to have an account, which is required to access Siri AI. For those younger, an account must be created by an adult, and these accounts can only access the basic, older version of Siri. When creating an Apple account, you must provide your date of birth, and many countries around the world are implementing stricter verification checks .
In most US states, anyone over 13 can have an Apple account and use the Siri AI app, with or without a linked parental account. This means that protections for teens are virtually identical to those for adults: there are measures aimed at preventing self-harm, suicide, sexual and violent content, and illegal commands. It is one of the most conservative AIs in terms of conversation topics.
Siri’s improved artificial intelligence can be controlled to some extent through the new parental controls available in iOS 27 and other Apple software updates for 2026—Apple clearly knows which way the wind is blowing—but for teenagers, this isn’t necessary. That depends on whether you can group your children into a Family Sharing group.
If your teens are part of your family group, standard Screen Time rules are used to control app usage limits and device access (in iOS Settings, tap your name, then select “Family” to access them). You can also disable some Siri AI features, such as writing tools or image creation.
Co-pilot
Like Gemini/Google and Apple/Siri, Copilot is integrated into the broader Microsoft ecosystem, and access for those under 18 can be controlled through it. Users must be over 13 to have a Microsoft or Copilot account—and again, the system uses technology to identify those who intentionally conceal their age.
If your teen uses Copilot, you can manage their access through the standard Microsoft Family Safety portal . You can enable or disable access to Copilot, set times when Copilot can and cannot be used, and disable certain features, such as image generation or voice mode. However, parents won’t be able to read their children’s Copilot chats.
Parental controls are optional—teens aged 13–17 can use Copilot without linking their account to an adult account. Microsoft states that users identified as under 18 will not see personalized results or advertising, but will receive age-appropriate responses regarding risky or harmful content, although details on what exactly this means are still scarce.
In general, the AI platform prohibits content related to self-harm, violence, adult content, harassment, and harmful or illegal activity. Any queries about suicide or drugs will offer support options rather than answers, regardless of whether you are an adult or a young person using the service.
Grock
Among major AI chatbots, Grok performs the worst in protecting minors. It can be accessed through X or as a standalone app, and although its terms of use state a minimum age of 13, a recent report from Common Sense Media claims that age verification on the platform is ineffective.
Ultimately, this means that minors have the same access to Grok as adults if they choose—an AI-powered platform that combines “weak security measures” and “a range of highly harmful content,” including “sexually explicit content, mental health risks, violent content, and inappropriate responses to… offensive content” (again, this is Common Sense Media’s report).
Yes, this is the same AI chatbot that previously had issues with pornographic images created using deepfake technology and has traditionally been rather lax about security, led by SpaceXAI CEO Elon Musk. While illegal and harmful content is supposedly prohibited on Grok, its own FAQ acknowledges that some chats may lead to “dialogue that may include profanity, crude humor, sexual situations, or violence.”
Grok Kids Mode has a special mode that limits the type of content displayed and topics of discussion, but it requires users to verify their underage status and is easily circumvented. Common Sense Media’s advice appears to be the optimal solution for Grok: don’t allow teenagers to access it.
Claude
Claude’s approach to child safety is simple: only those over 18 can register an account. You can hide your age during registration, but like other platforms, Anthropic monitors various behavioral “signals” to identify minors. If your account is flagged as suspicious, you’ll be required to verify your age using tools like facial recognition or Yoti.
Unlike ChatGPT and Gemini, there is no free web access to the chatbot, so if you are under 18, access to the service will be blocked unless you know behavioral tricks to bypass Claude’s detection systems.
As for more general well-being protections on the Claude platform, any discussions related to self-harm or suicide will direct users to hotlines and other resources for support. Discussions about illegal activities, prohibited speech, and sexually explicit content are prohibited on the platform. In recent months, Anthropic also reports that it has reduced the level of AI sycophancy.
This is a brief overview of the seven largest AI-powered chatbots, but if you’re a parent, it should give you an idea of what’s out there and how easy it is to access. In addition to the built-in (or non-built-in) parental controls in these AIs, you also have platform-level settings—Android, iOS, Windows, macOS—to restrict the use of any potentially harmful apps.