APPLIED GUIDE

What not to share with ChatGPT or other AI tools

Do not place data into an AI tool merely because it would make the prompt easier. Check tool approval, purpose, retention and access terms, then remove identifiers and sensitive details. When the task can be completed with synthetic, aggregated, or redacted data, use that safer form.

Responsible authorVidAcademia Editorial Team
ReviewVidAcademia curriculum and safety review
MethodPrimary-source synthesis, applied scenario, verification checklist, and human review.

Recommended process

Do not place data into an AI tool merely because it would make the prompt easier. Check tool approval, purpose, retention and access terms, then remove identifiers and sensitive details. When the task can be completed with synthetic, aggregated, or redacted data, use that safer form.

  • Do not enter passwords, secrets, tokens, or private keys.
  • Avoid personal records and identifiable student, health, financial, or employment data.
  • Do not upload confidential contracts, source code, or customer data without approval.

Review checklist

Use this checklist before accepting the output or turning it into an action.

  • Remove names, identifiers, and unnecessary context.
  • Use an approved enterprise tool and policy when sensitive work is authorized.

CONCRETE EXAMPLE

Observable result

Instead of pasting a customer complaint with names and account numbers, a team uses a redacted scenario that preserves the issue type and desired response.

  1. Do not enter passwords, secrets, tokens, or private keys.
  2. Avoid personal records and identifiable student, health, financial, or employment data.
  3. Do not upload confidential contracts, source code, or customer data without approval.
The work remains traceable, reviewable, and tied to a human decision.

PRIMARY SOURCES

Check the basis for this guide.

Frequently asked questions

Does deleting a chat remove every risk?

No. Prevention is safer: understand the service terms and avoid entering data that is not approved for that system.

Is a citation enough to trust an answer?

No. Confirm that the cited source exists, is current, and actually supports the claim made.

AI Literacy at Work: Prompt, Verify, Protect

Use generative AI for a real work task with a repeatable verification and privacy checklist.

View the free course

Continue this topic

Human review in AI workflowsHow to evaluate AI output quality