Privacy score: 50/100. Transparency score: 60/100. Overall 52/100 (Not Recommended), reviewed 2026-06-29.
What the vendor documentation says
OpenAI's privacy policy (updated April 28, 2026) states that conversation content on the consumer tiers is used to improve and train models by default. Users must actively navigate Settings to disable training. OpenAI does not accept users under 13 but relies entirely on self-declared birthdates with no verification. The April 2026 policy retains broad data retention rights, and third-party integrations can receive conversation context. The 13+ age threshold places ChatGPT outside COPPA's direct scope, leaving a regulatory gap for 13-15-year-olds.
Controls and safeguards on the data path
OpenAI launched parental controls September 29, 2025, allowing parents to link teen accounts, set time limits, restrict content, and receive notifications. A safety-routing system routes sensitive conversations to stronger reasoning models. Age prediction was announced September 16, 2025. Key gaps remain: parental controls require both parent and teen to opt in and link accounts; the platform has no hard age verification at signup. Persona and memory features can create parasocial attachment patterns in vulnerable teens.
Open privacy concerns
- At least five active wrongful-death lawsuits allege ChatGPT facilitated teen suicides
- OpenAI's legal defense blamed a deceased teen for 'violating Terms of Service'
- No hard age verification at signup
- Parental controls are opt-in and require mutual account linking
- Conversations used for model training by default; opt-out is buried
- Persona and memory features may foster parasocial attachment
- Consumer product not FERPA/COPPA compliant
- No published independent third-party safety audits
Questions to ask before you sign
- Does the license we are buying include a signed FERPA data processing agreement, or is it a consumer subscription?
- Is training on our users' conversations disabled contractually, not just by a settings toggle?
- What is the retention period for conversation logs, and can we request deletion?
- Is there an admin console with per-user provisioning and audit logs?
- Has any independent third party audited the child-safety behaviour of the model version we would receive?
District posture we recommend
School IT administrators should not deploy consumer ChatGPT Free or Plus for student use. These tiers do not provide FERPA DPAs, admin dashboards, or COPPA-aligned data handling. OpenAI offers ChatGPT Edu as a separate institutional product. Multiple states (Ohio with a July 1, 2026 deadline) now mandate that districts adopt formal AI policies; the consumer tiers are unlikely to meet compliance requirements.
What this article does not claim
Findings labelled as verified come from SafeGradeAI's published evaluation of this tool and the vendor's own public documentation as of the review date above. Anything framed as a recommendation is our 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.
Frequently asked questions
Is ChatGPT safe for a 10-year-old?+
No. OpenAI's own terms set a 13+ minimum and SafeGradeAI rates ChatGPT not recommended for ages 5-8 and 9-12. There is no child mode on the consumer tiers.
Do OpenAI's parental controls actually stop unsafe conversations?+
They add time limits, content restrictions, and notifications, but they are opt-in and require both parent and teen to link accounts. They do not add age verification, and a teen can create an unlinked account with a different email.
Is ChatGPT allowed in schools?+
Consumer ChatGPT Free and Plus do not come with a FERPA data processing agreement or an admin console, so SafeGradeAI does not recommend them for student use. ChatGPT Edu is a separate institutional product with different terms.
Are our conversations used to train the model?+
On the consumer tiers, yes, by default. You can opt out in Settings > Data Controls; the setting is not on by default.
Recommended next steps
For schools
Govern AI tools district-wide with School Pro
Unlimited full evaluations plus the K-12 workspace — tool registry, approval workflow, monitoring, and a downloadable audit log.
For parents
Start with the Family AI Safety Starter Pack
A free, practical pack you can use alongside ChatGPT — conversation starters, age guidance, and a family AI agreement.
Want the deep dive? Get the Complete AI Safety Blueprint — $29