Is ChatGPT Safe for Kids?

SafeGradeAI verdict: Not Recommended — 52/100. Last reviewed 2026-06-29. Fact-checked against OpenAI's current documentation on 2026-08-24.

The short answer

ChatGPT (consumer Free and Plus tiers) is OpenAI's flagship general-purpose AI chatbot. OpenAI's own help centre says ChatGPT is not meant for children under 13, that 13-17 year olds need a parent or guardian's permission, and that any under-13 educational use must be conducted by an adult. Since August 18, 2026 OpenAI has also placed users whose declared or estimated age is 13-17 into ChatGPT for Teens, a separate teen experience with stronger default protections. That is a real improvement — but it arrived after a series of safety failures involving minors, age is still largely self-reported, and the consumer tiers were designed for adults. SafeGradeAI does not recommend the consumer product for children, and recommends teen use only with permission, linked parental controls, and household rules.

Age-by-age guidance

| Age band | SafeGradeAI guidance | | --- | --- | | Ages 5-8 | Not recommended | | Ages 9-12 | Not recommended | | Ages 13-18 | Use with caution (teen experience + parental controls) |

How it scored

| Pillar | Score | | --- | --- | | Privacy & data protection | 50/100 | | Content moderation | 46/100 | | Safeguards & controls | 56/100 | | Age-appropriate design | 52/100 | | Transparency | 60/100 | | Overall | 52/100 — Not Recommended |

What the vendor documents (verified)

• Minimum age is 13, or the local digital-consent age where that is higher; under-18s need parent or guardian permission (OpenAI Terms of Use).
• ChatGPT for Teens (launched August 18, 2026) automatically applies a teen experience to users whose age is declared or estimated as 13-17.
• Parental controls let a linked parent manage selected settings and quiet hours. They do not let a parent read the teen's conversations.
• Consumer chats may be used to improve models depending on settings. "Improve the model for everyone" can be switched off in Data Controls; chats can still remain in history. Temporary Chat is not used for training.
• ChatGPT Business, Enterprise and Edu inputs and outputs are not used to train models by default.
• OpenAI states that ChatGPT can produce incorrect facts and fabricated citations, and that important claims should be verified.
• OpenAI states that AI-content detectors have not been reliable enough for high-stakes judgements about students.

Where the risks are (SafeGradeAI assessment)

• Age assurance is still largely self-reported or estimated; a determined teen or pre-teen can present as an adult.
• The teen experience and parental controls are only as good as the account linkage behind them — an unlinked second account bypasses both.
• Moderation is generic rather than curriculum- or child-tuned, and can be circumvented by framing.
• The consumer tiers carry no FERPA data processing agreement or admin console, so they are not appropriate for student data.
• Memory and persona features can encourage parasocial reliance in vulnerable teens.
• SafeGradeAI's low moderation score reflects the documented 2025 safety failures involving minors in acute distress and the reactive pattern of the fixes that followed.

Parent & educator checklist

• Is the student at least 13 (or the local digital-consent age), and do they have your permission?
• Is the account the student's own, under their real age, so the teen experience actually applies?
• Have you linked parental controls and set quiet hours — while being clear that you cannot read their chats?
• Have you agreed what must never be typed in: full name, address, school, logins, medical or financial detail, photos?
• Does the student understand that ChatGPT can be confidently wrong and can invent citations?
• Is AI being used to support learning, not to complete assignments?
• Have you read your school's AI policy, and does the household rule defer to it for schoolwork?
• Do you review the arrangement together on a set schedule, openly rather than covertly?

Our recommendation

For ages 5-12: not recommended. For 13-17: allow only with your permission, on the student's own correctly-aged account so the teen experience applies, with parental controls linked, "Improve the model for everyone" switched off, and a written household rule set. Schools should not deploy the consumer tiers with student data; ChatGPT Edu is a separate institutional product with different terms and a different review path.

What this article does not claim

Findings labelled as vendor-documented come from OpenAI's own public documentation, linked in Sources below and last checked on the date shown. Anything framed as an assessment or recommendation is SafeGradeAI's professional guidance, not a statement of fact about the product. Vendor policies, model versions, and default settings change frequently — re-check the vendor's current documentation before making a district-level decision, and treat any AI output as a draft to verify.

---

This guide is built on the SafeGradeAI evaluation of ChatGPT. Read the full ChatGPT evaluation and rubric scores. See also the ChatGPT parent guide, privacy guide, age guide and family rules.