Family Check
Explore AI chatbots together with your child. Talk about which answers seem helpful and which seem wrong, and review privacy settings together.
UNICEF, Parenting in the AI ageWatch for warning signs: secrecy, anxiety, relying on AI for emotional support, or AI crowding out sleep, school and friends. Ask gentle questions, then agree simple limits.
UNICEF, Parenting in the AI ageKeep social AI companion apps away from under-18s. Common Sense Media rates them an unacceptable risk and says they are designed to create emotional attachment and dependency.
Common Sense Media, Social AI Companions risk assessmentAgree a family safe word, set social media accounts to private, and limit public photos and voice clips that scammers could copy.
FBI IC3, Criminals Use Generative Artificial Intelligence to Facilitate Financial FraudIf someone threatens to share a sexual image of your child, real or fake: don't pay, keep the evidence, report it, and use NCMEC's Take It Down.
FBI IC3, Malicious Actors Manipulating Photos and Videos to Create Explicit Content and Sextortion SchemesTick what your family already does. Your ticks stay on this page only; we don’t save them.
MostUS teenagers use AI chatbots.Pew Research Center, February 2026 (1,458 US teens)
72%of US teens have used AI companions, an AI “friend”.Common Sense Media, July 2025
Rules that protect children from AI chatbots and companions. They differ from place to place, and some start later.
Bans AI that exploits age-related vulnerabilities to distort behaviour and cause significant harm. New bans cover AI that makes child sexual abuse material or non-consensual sexual deepfakes ('nudifier' apps).
Where it standsAge-vulnerability ban applies since 2 Feb 2025; new bans apply from 2 Dec 2026.
Official textAI systems that talk with people, such as chatbots, must tell users they are dealing with an AI unless it is obvious. Deepfakes must be clearly labelled.
Where it standsApplies from 2 Aug 2026. AI systems already on the market before then have until 2 Dec 2026 to mark AI-made content.
Official textPlatforms used by children should set private accounts by default, switch off features like streaks and autoplay, use age assurance, and add safeguards to built-in AI chatbots.
Where it standsPublished 14 Jul 2025; the Commission uses them to check compliance.
Official textWould require AI chatbots and companions to be off by default for minors and ban them from making children emotionally dependent.
Where it standsProposed 17 Sep 2026; not yet law.
Official textOperators must stop chatbots exposing minors to sexual content, remind minors they are talking to AI, warn chatbots may not suit minors, and refer users with suicidal thoughts to crisis services.
Where it standsIn force since 1 Jan 2026.
Official textAdds age assurance, risk tests before release, parent-only default settings, crisis referrals, parent alerts on self-harm, and independent audits for companion chatbots used by children.
Where it standsSigned 10 Sep 2026; not yet in force.
Official textAI companions must detect signs of suicidal thoughts or self-harm and refer users to crisis services, and remind users every three hours they are not human. Applies to all users, not only children.
Where it standsIn force since 5 Nov 2025.
Official textThe lawPlatforms must remove intimate images shared without consent, including AI-made fakes, within 48 hours of a valid request.
Where it standsSigned 19 May 2025; FTC enforcing platform duties since 19 May 2026.
Official textRisky sites must use strong age checks and keep children from porn, self-harm, suicide and eating-disorder content. Feeds must not push harmful content to children.
Where it standsIn force since 25 Jul 2025.
Official textMost of the chatbots that children use most are already covered. Some are not, because of the way they work, and the government says it will close this gap. It plans to keep under-18s from chatbots that mainly offer sexual content, and to make chatbots give under-18s breaks.
Where it standsGovernment response published Jul 2026; not law yet.
Official textCommon Sense Media AI risk assessments (Youth AI Safety Institute) tests AI products used by kids and teens, and rates each one.
Four levels, from low to unacceptable risk.
All their ratings| App | Risk | Updated |
|---|---|---|
| ChatGPT for Teens | unacceptable | 7 Oct 2026 |
| Perplexity | unacceptable | 15 Sep 2026 |
| Google AI search | unacceptable | 14 Jul 2026 |