Responsible AI Checklist

How to use this checklist

You do not need every item for every task. Asking AI to rephrase a sentence is low stakes; using it for a client report, a grade, or a medical question is not. Run the full list whenever the output will affect someone else or carry your name. The habit takes about a minute and prevents the failures that actually get people in trouble.

The checklist

Verify before you rely on it

• Verify names, numbers, dates, and quotes against a real source before sharing.
• Ask for sources, then actually open them — AI tools sometimes cite pages that do not exist or do not say what is claimed.
• Treat confident tone as meaningless; models sound equally sure when they are wrong.
• For anything high-stakes (legal, medical, financial), use AI to prepare questions, not to replace a qualified professional.
• Keep a habit from How to Verify AI Answers: the more surprising the claim, the harder you check it.

Protect private information

• Never paste confidential work data — client names, financials, unreleased plans, internal documents — into a public AI tool.
• Strip personal details (yours or anyone else's) from prompts: full names, addresses, ID numbers, health information.
• Assume anything you type may be stored; if you would not email it to a stranger, do not prompt with it.
• Check whether your school or employer has an approved AI tool and a usage policy before using your personal account for their work.

Disclose and attribute

• Follow your school's or employer's rules on AI use — and when the rules are unclear, ask before submitting, not after.
• Tell people when substantial parts of a deliverable were AI-generated; getting caught hiding it costs more than disclosing it.
• Do not present AI output as your own analysis in contexts where original work is the point.
• Remember that you own the result: "the AI wrote it" is not a defense for errors or plagiarism.

Watch for bias

• Notice who is missing: AI summaries and recommendations can quietly favor majority perspectives from their training data.
• Be extra careful when AI touches decisions about people — hiring, grading, lending, moderation — and keep a human review in the loop.
• Ask for the opposing view or the strongest counterargument before accepting a one-sided answer.
• If an output stereotypes a group, do not reuse it, even lightly edited.

Use good judgment

• Match the tool to the stakes: drafting and brainstorming are great uses; final decisions about people or money are not.
• Do not use AI to deceive — fake reviews, impersonation, or manufactured evidence are misuse regardless of the tool.
• Keep your own skills sharp; if you cannot evaluate the output, you are not ready to delegate the task.
• When something feels off, stop and check with a person. Discomfort is data.

Make it a habit

Responsible use is a routine, not a rulebook. Pin this list next to the Prompt Engineering Cheat Sheet, and if you use AI on the job, the AI at Work learning path covers how to build these habits into daily workflows.