Module 3: Knowing what not to trust — Knowing what NOT to trust is huge. We'll learn how to spot when AI makes things up.
- Scenario: A student wants to paste their entire medical history into ChatGPT to get health advice. Is this safe?
- No — medical records are personal health data and should never be shared with a public AI. — Health data is among the 5 things to never paste. Apply the billboard test.
- Scenario: Before pasting a work document into AI, what should you always check first?
- Whether your company has an AI usage policy that covers what you can and cannot share. — Company policies vary widely — some ban AI usage on internal docs entirely.
- Scenario: A freelancer wants to analyze client revenue data using AI. What's the safest approach?
- Anonymize the data first — replace company names and exact figures with placeholders, then paste it. — The AI doesn't need real names or actual numbers to help you analyze patterns.
- Scenario: You're on a free AI plan. You notice your chat settings say 'conversations may be used to improve our models.' What should you do?
- Turn off the training data toggle in settings — your conversations may otherwise be used for model training. — Most free plans let you opt out of training in the settings. You just have to find it.
- Which of these is safe to paste into any AI without concerns?
- A publicly available Wikipedia article you want summarized. — If the information is already public, there's no privacy risk in sharing it.
- Scenario: A manager wants to paste employee performance reviews into AI to help write promotion recommendations. What's wrong with this?
- Performance reviews contain private employee data that shouldn't be shared with third-party AI tools. — Employee reviews contain PII, opinions, and sensitive HR data — all on the 'never paste' list.