Read the handout here
The full handout is available below in web text. You do not need to enter an email to read it.
1. Before you open a tool
Choose the work: What are you trying to do? Could a person, template, search or ordinary software do it better?
Check approval: Use only tools, accounts and information your organisation has approved. If you are unsure, ask your manager or the named AI or data lead.
Keep information out: Do not enter passwords, customer or staff details, health or financial information, confidential plans, private documents or material you do not have permission to share.
2. Give a clear instruction
State the task, useful fictional or approved context, the audience and the format you want. Set boundaries: say what the tool must not assume or invent. Ask it to flag uncertainty.
Practice example: “Using only this fictional event note, draft a five-step room set-up checklist. Do not invent times or names. Mark anything a person must confirm.”
3. Check before using the answer
Read the full output. Verify facts against a trusted source. Look for missing context, unfair assumptions, made-up references and unsafe advice. A confident tone is not proof. Get the right person to approve anything important.
4. Pause if something feels wrong
A voice note, image, email or video can be faked. For an unexpected request, especially one about money, access or personal details, stop and verify through a separate trusted channel. Follow your organisation’s reporting route.
5. Learn what actually helped
Note the starting task, what changed and what still needed human work. Compare quality and effort, not only speed. Share useful learning without sharing confidential prompts or judging individual staff by tool activity.
The human-first rule: If you cannot explain or check the result, do not rely on it.
General learning guidance, not legal, security or professional advice. Your employer’s approved tools and policies take priority.