Protect data and rights
Use AI without leaking sensitive information or ignoring ownership and policy constraints.
AI and Machine Learning with Google Cloud
From Google Cloud. Watch here or open it on YouTube .
The prompting techniques are only appropriate after data handling is approved. Identify which example inputs you would need to redact in a real workplace.What this lesson is really solving.
Assume inputs may be retained or processed under terms you have not memorized. Minimize data, remove identifiers, follow approved-tool policies, and never paste secrets. Check ownership, licensing, attribution, confidentiality, and affected-person expectations before publishing outputs.
Understand the idea before touching the steps.
Data risk depends on what enters, where it goes, how long it stays, and who can reuse or access it.
Follow these steps in order.
Take the action in each step; then use the deliverable below to prove the lesson is finished.
- 1
Classify data before use: public, internal, confidential, regulated, personal, or contract-restricted.
- 2
Read the tool’s current retention, training, access-control, and enterprise settings instead of assuming privacy.
- 3
Minimize inputs by redacting names, identifiers, secrets, customer data, and unnecessary source material.
- 4
Record permitted tools, prohibited data, review responsibility, and escalation steps in a simple policy.
See the standard in context.
Instead of uploading customer tickets, an analyst creates synthetic examples with the same categories. The team uses an approved enterprise workspace, disables optional sharing where available, prohibits API keys and health data, and requires human review for messages sent to customers.
Inspect before you move on.
- No password, token, private key, or authentication code enters a prompt.
- Sensitive fields are removed or replaced before processing.
- Tool terms, organizational policy, and rights to source material fit the intended use.
Your deliverable
A one-page AI use policy and redaction checklist for your actual environment.
Watch for this
Copying a full confidential document when only two nonsensitive paragraphs are necessary.
Prove it—don’t just recognize it.
You can decide whether a given input is allowed before opening the AI tool.
Quick knowledge check
Answer from the lesson—not from confidence alone. Score at least 2 of 3 to unlock completion.
This curriculum-aligned check is scored automatically and stored with your account when signed in. It is an objective learning signal, but it has not yet been independently validated as a standardized assessment.
Tools, templates, and references
Pass the knowledge check above first.
Completion unlocks after a score of 2 out of 3. Then confirm that you produced the lesson deliverable.
Go to the knowledge check