No. ChatGPT and Meta AI give parents a list of topics, or an alert in a crisis, but neither shows you the conversation. Meta will show you the topic your teen asked about even when its AI refused to answer.
I do get why. A teenager who knew a parent could read every chat would simply stop being honest with it. What worries me is that the alerts only come once something is already serious.
ChatGPT, made by OpenAI. Since 18 August 2026, anyone ChatGPT estimates is under 18, or who gives an age between 13 and 17, is moved automatically into ChatGPT for Teens.
If you link your account to your teen’s, you can set quiet hours when ChatGPT can’t be used, and switch off voice, image generation and memory, so it stops building up a record of everything your teen has told it. Linking also keeps extra safeguards on against sexual or romantic role-play, graphic content and extreme beauty ideals.
You cannot read the chats. You are notified only when trained reviewers spot a serious self-harm concern or signs of disordered eating, or when the account is banned for conversations about violence, and the notification doesn’t include what was said. Your teen can unlink whenever they like, and you are told if they do.
Meta AI, the assistant inside Instagram, Facebook and Messenger. With supervision on, you see the topics your teen asked about over the past seven days, such as school, entertainment, or health and wellbeing, broken down further into areas like fitness or mental health. You don’t see the questions or the answers.
Since 5 October 2026 the same supervision works in the separate Meta AI app, worldwide. Since 29 October, supervising parents everywhere get an alert if a chat with Meta AI suggests a risk of suicide or self-harm, and a person reviews each flagged chat before the alert goes out. A stricter setting, Limited Content, makes Meta AI turn down more requests. There’s no off switch for Meta AI inside Instagram, which tells you how keen Meta is for teens to use it.
Character.AI, the role-play chatbot app. Since November 2025, under-18s can’t have open-ended chats with its characters, and get story and video features instead.
When the company that built a companion chatbot decides its own product isn’t safe for under-18s to chat with freely, I’d take the hint.
Everything else. A chatbot with no parent account gives you nothing. And every teen version above only applies once the company has decided the user is a teen.
The law is moving, but not towards transcripts. New York signed a law on 28 October 2026, in force early next year, that bars chatbots offered to children from pretending to be human, building emotional dependency, encouraging self-harm or having sexual conversations. The EU has proposed switching AI companions off by default for under-18s. Neither gives parents the conversation.
OpenAI and Meta have both changed these controls more than once since July. Subscribers hear the next time they do.
Every Thursday we answer the questions parents are really asking. Free.
What this means for you right now
If your child uses a chatbot, this is what I’d do.
Link the accounts anyway. It is the only way to get the crisis alerts, and on ChatGPT it switches on stricter safeguards. Set it up with your teen sitting next to you. They can undo it, and you’ll be told if they do, so it only works if they’re on board.
Treat the topic list as a reason to ask. A week of “health and wellbeing” is a good reason for a chat in the car, where nobody has to make eye contact.
Ask which chatbots your teen uses. These controls cover two companies. A chatbot inside a game, a browser or another app is outside all of them. “Which AI do you use?” gets a better answer than “Show me your phone.”
Don’t treat the alerts as a safety net. OpenAI says its notifications aren’t real-time monitoring and may miss things. Meta’s only reach supervised accounts.
Send this to another parent on WhatsApp
Find out how the AI apps rate at each age is in the scorecard.
Checked 2 November 2026.
Sources:
- OpenAI, Introducing ChatGPT for Teens, 18 August 2026
- OpenAI Help Center, parental controls in ChatGPT
- Meta Newsroom, parental insights into teens’ conversations with AI, updated 5 October 2026
- Meta Newsroom, alerting parents to signs of distress in conversations with Meta AI, updated 29 October 2026
- Transparency Coalition, New York’s chatbot safety law signed, 28 October 2026
