Start by checking eligibility and the settings visible in the actual account. Then agree on a learning task whose result can be checked. An enabled setting does not prove that a student understood an answer.
Updated September 24, 2026: this replaces the launch-focused article with current documentation and a reusable study exercise. It does not evaluate the effectiveness of OpenAI’s safeguards.
Check eligibility before setting up a study routine
OpenAI’s age guidance says ChatGPT is not intended for children under 13 and requires parental consent for ages 13–18. It also cautions that output can be inappropriate for a classroom. Follow your school’s rules and applicable local requirements before using it with students.
The teen-experience documentation describes automatic enablement for eligible accounts identified as under 18. Do not assume a teacher’s demonstration account has the same settings as a student’s account.
Separate access, study and privacy choices
According to OpenAI’s parental-control documentation, a parent or guardian can invite a teen through Settings → Parental controls; the connection requires acceptance. Linked controls do not provide access to chat transcripts. Either person can unlink the accounts.
Study hours make eligible new chats start in study mode during a schedule; quiet hours restrict access during selected times. Study hours do not apply to GPT or Project conversations. Settings may affect new chats without changing existing ones. Safety notifications cover limited situations and are not real-time monitoring or a replacement for professional care.
For a family review, record three decisions in your own words:
| Decision | Record before the first task |
|---|---|
| Allowed work | Which assignment or practice task can use AI under school rules? |
| Information shared | Which public or fictional input will be used? What should stay out? |
| Evidence of learning | What must the learner explain or correct without copying the answer? |
This worksheet is an AIViewer teaching suggestion, not a product feature or a validated safety intervention. Keep the conversation about these choices separate from promises about what a filter can prevent.
Try a short exercise with a checkable source
The following notice is fictional teaching material:
The library club meets on Tuesday from 15:30 to 16:15 in Room B. Bring a notebook. The club is free. Sign up with the librarian by Monday. This notice does not say whether snacks are provided.
Ask the learner to identify the meeting duration, cost and missing information before using an assistant. A suitable instruction is:
Help me understand this notice. Ask one question at a time. Let me calculate the duration first. Keep facts from the notice separate from guesses, and do not write a finished answer for me to submit.
The instruction is a proposed prompt. AIViewer has not tested it on a teen account, and it does not guarantee how a model will respond.
Check a deliberately flawed answer
Here is an authored incorrect answer, not an observed chatbot output:
The club lasts one hour, costs 5 dollars, and includes snacks. Sign up on Tuesday.
Use the notice as the answer key:
- From 15:30 to 16:00 is 30 minutes; another 15 minutes makes 45 minutes, not one hour.
- The stated cost is free. The 5-dollar fee was invented.
- Snacks are unknown. Missing information is not permission to guess.
- Sign-up is due Monday, even though the meeting is Tuesday.
A corrected response is: “The free club meets Tuesday for 45 minutes. Sign up by Monday. Ask the librarian whether snacks are provided.” Each part has a visible basis or identifies a question that remains open.
Check understanding with a changed input
Change the finishing time to 16:05. The new duration is 35 minutes. Ask the learner to explain the subtraction and identify which other facts remain unchanged. The cost, room and sign-up deadline should not change merely because the time changed.
If the learner can only repeat the previous answer, return to the notice and work through the times together. This is a practical way to inspect one piece of reasoning; it is not a claim that using AI improves learning overall.
For real assignments, retain the source, the learner’s own working and any required AI-use disclosure. Use the claim-checking worksheet when a response mixes facts with unsupported additions.