The AI Ethics Checklist for Project Managers
Most AI ethics material is designed for organizations: policy documents and committee debates about industry-wide issues. That's legitimate work, and it rarely helps in the moment you're actually deciding whether a specific line belongs in a specific prompt.
Seven questions, one for each of the Seven Deadly Sins of GenAI. Answer them before AI-assisted work leaves your hands.
Free. Open to everyone. No email required.
"I Honestly Don't Know What's Safe to Paste Into This Thing"
You're using AI on real project work now, more than you were a year ago, and somewhere in that stretch you stopped asking whether you should.
Start with the non-negotiable list. Never put into a public AI tool:
- Personally Identifiable Information: names, emails, addresses, Social Security numbers
- Business Identifiable Information: proprietary data, financials, trade secrets
- Confidential project information that could damage your organization if it leaked
Check your organization's AI policy too. That list covers the floor, the baseline that holds no matter what your policy says.
Everything above that floor is a judgment call, the kind no policy is written to make for you. That's what the seven questions below are for.
What's Inside
Seven questions, one for each of the Seven Deadly Sins of GenAI, built to run through before you hit send:
| Sin | Category | The Question |
|---|---|---|
| Pride | Overconfidence | Have I verified this, or am I trusting it because it sounds sure of itself? |
| Greed | Data and Privacy | Would I be comfortable if what I just typed became public tomorrow? |
| Lust | Dependency | Did I think first and use AI second, or did AI become the only move I made? |
| Envy | Integrity and Attribution | Does the person receiving this know how much of it AI produced? |
| Gluttony | Excess and Waste | Did I stop once my purpose was met, or am I still generating "just in case"? |
| Wrath | Responsibility and Harm | Am I willing to put my name on this, fully, if it turns out wrong? |
| Sloth | Judgment | Could I still do this without AI, and have I actually checked lately? |
That's the one-page version: seven questions on a single sheet you can keep open next to whatever AI tool you use.
The download below also includes the Workbook Edition, an 8-page expanded version. Title page plus one page per sin, and each page gives you room to work the question instead of reading it: where that sin shows up in real PM work, a reflection field for your own answer, and one commitment to carry forward.
Built for You, Not Your Organization
Most AI ethics material is designed for organizations: policy documents and committee debates about industry-wide issues. That's legitimate work, and it rarely helps in the moment you're actually deciding whether a specific line belongs in a specific prompt.
This checklist is scoped differently. Four of the seven questions (Greed, Lust, Envy, and Wrath) are framed around what you do with the tool in your hands, your own practice, rather than what the AI industry should be doing differently. Pride, Gluttony, and Sloth read the same either way; the personal read and the industry read land in the same place.
The Workbook Edition builds toward one more artifact past the seven pages: a Personal AI Ethics Charter, a document meant to describe your own standard. A policy committee can write guidance for a whole organization. Only you can write that one.
Who It's For
Project managers already running real work through AI, who've never sat down and actually decided what their own standard is. Right now you default to whatever feels right in the moment. Some days that's fine. Some days you notice you can't quite say why.
Not for you if you're looking for an organizational AI policy, a legal compliance document, or a debate about what AI companies owe the people whose data trained them. That's real work, and it belongs in a different document, written for a different reader.
A Taste of It: Pride
Page 2 of the Workbook Edition, in full:
The question Have I verified this, or am I trusting it because it sounds sure of itself?
Where this shows up
- You paste an AI-drafted risk register into a client deck without checking a line of it against your own judgment.
- AI-generated status language stands in for your own read on the project, because it read confidently.
- You stop asking whether it's actually right once the output sounds polished.
Carrying this forward Treat AI as a first draft, not a final authority. Build a review step you can't skip.
Six more pages like this one follow, each with its own reflection field. The one-pager gives you the question. The workbook gives you room to actually answer it.
Get the Checklist
Free. Open to everyone. No email required.
Two downloads, two documents: the one-page checklist and the eight-page Workbook Edition.
Ready to Build Your Own Charter?
If you want additional help building that Personal AI Ethics Charter, I built a tool to help you.
If you wish to see my own output from the tool, you can see my post here.
And if you want to see my reasoning about why I constructed this around the Seven Deadly Sin, you can read that here.
Want one idea like this every week? The Intentional PM newsletter: weekly insights for project managers who lead with AI.