Purpose

Use this lesson to decide what must be removed before using an AI tool. The aim is to produce work that a colleague can review and act on, not merely text that sounds convincing.

Lesson Explanation

Scenario: An employee wants AI help summarizing a document containing customer addresses and contract prices.

Better instruction: Create a safe workflow: remove personal identifiers and confidential prices, replace them with placeholders, summarize only the approved text, and have a manager review the output before sharing.

Before prompting, classify the information: public, internal, confidential, or personal. Minimize data, anonymize where possible, follow your organization’s approved tools and policies, and keep a human accountable for disclosure decisions. Addresses and prices are removed or replaced before the approved text is summarized.

Work in a short cycle: provide the relevant input, state the result and limits, inspect the draft against the acceptance criteria, then revise the instruction when a requirement is missing. Keep the final human decision with the person responsible for the work.

Practice Questions

1. An employee wants to use an AI tool to summarize a document containing customer addresses and confidential prices. What does this lesson say should happen before prompting?

   View Answer

   The information should be classified: public, internal, confidential, or personal – before deciding how to proceed.

2. What does this lesson recommend doing with personal identifiers and confidential data before summarizing a document with AI?

   View Answer

   Removing or replacing them with placeholders, then summarizing only the approved text.

3. Write a prompt-and-process instruction for safely summarizing a document with sensitive customer data, following this lesson’s guidance.

   View Answer

   Something like: “Remove personal identifiers and confidential prices from the document, replace them with placeholders, summarize only the remaining approved text, and have a manager review the output before sharing.”

4. An employee pastes an entire confidential contract, including client names and prices, directly into a general AI tool without any review. What does this lesson say about this practice?

   View Answer

   This skips classifying the information and minimizing/anonymizing it, both of which this lesson recommends before using AI on sensitive material.

5. What does this lesson say should happen before an AI-summarized document containing originally sensitive information is shared?

   View Answer

   Manager review, keeping a human accountable for disclosure decisions.

6. A workflow replaces customer names with “Customer A,” “Customer B,” etc., before submitting a document for AI summarization. What principle does this reflect?

   View Answer

   Anonymizing sensitive information before processing, consistent with this lesson’s guidance.

7. A completed safe-summarization process removes personal identifiers and prices, replaces them with placeholders, summarizes only approved text, and includes manager review. What does this reflect?

   View Answer

   A complete, safe workflow following this lesson’s full guidance.

8. Why does this lesson recommend using “your organization’s approved tools and policies,” rather than any available AI tool?

   View Answer

   Different tools may have different data handling and retention practices, so using approved tools ensures the organization’s privacy and security standards are actually being followed.

9. What four categories does this lesson recommend for classifying information before using AI on it?

   View Answer

   Public, internal, confidential, or personal.

10. A team wants to minimize the amount of sensitive data sent to an AI tool at all. What principle from this lesson does “data minimization” reflect?

    View Answer

    Minimizing data as one of the core practices this lesson recommends, alongside anonymizing and following approved tools/policies.

11. Why does this lesson emphasize keeping “a human accountable for disclosure decisions,” even when AI assists with the summarization itself?

    View Answer

    Because deciding what sensitive information can appropriately be disclosed or shared remains a human responsibility, not something to be delegated entirely to an AI process.

12. An employee classifies a document as “confidential” before deciding how to handle it with an AI tool. What does this reflect, per this lesson’s guidance?

    View Answer

    The first recommended step: classifying the information before determining how it should be processed.