AI Policy Builder

1What’s your organization called?

Name as it should appear in the policy

2What kind of organization is it?

This sets the examples, the vocabulary and a section written for your field.

Museum, gallery or cultural organization

Library or archive

Food bank or human services

Professional or member association

College, school or education program

Foundation or grantmaker

Another kind of nonprofit

3How many people work there, counting part-time staff?

Fewer than 10

10–50

51–250

More than 250

4How much are staff using tools like ChatGPT, Claude or Copilot today?

Your best guess is fine. Most leaders guess low.

Hardly at all, as far as we know

A few people, on their own

Lots of people, mostly on personal or free accounts

Lots of people, on accounts the organization pays for

5Which AI tools does the organization provide or pay for?

Tick all that apply. These become your approved tools.

ChatGPT, on an organization plan

Claude, on an organization plan

Microsoft Copilot, through work Microsoft 365 accounts

Gemini, through work Google accounts

AI features inside software we already useDonor database, Zoom, Canva, email and the like

None yet

6What private information do you hold?

Tick all that apply. Each one gets named in the policy, so staff know exactly what to keep out of the wrong tools.

7Which rules do you already have to follow?

Tick any you know apply. “Not sure” adds a question for your lawyer to the draft.

HIPAA, for health information

FERPA, for student records

GDPR, for people in Europe or the UK

A US state privacy law, such as California’s

A state law protecting library patron records

Terms in government grants or contracts

Not sure

8What do you produce that the public sees?

Tick all that apply.

Newsletters, social posts, web pages, appeals

Grant applications and reports to funders

Exhibition text, labels, catalogs or other publications

Classes, trainings or education materials

Research, journals, standards or guidance for a field

Materials in more than one language

9How do you feel about AI-made images in things you publish?

Not in anything we publish

Fine for drafts and mock-ups, not for publication

Fine in public, if clearly labeled

Fine, as long as someone reviews them

10Does any team object to using AI in their work?

It’s common. Archivists, clinicians and educators often have good reasons.

Yes, at least one team does

Some individuals, not whole teams

Not that we know of

Which team? (optional)

11Does AI play any part in decisions about people?

Hiring, evaluations, grant awards, who gets services, admissions, membership.

No, and we’d like to keep it that way

It might help sort or summarize, with people deciding

Not sure

12Will the public talk to an AI on your behalf?

A chatbot on your website, a kiosk, an AI phone line.

No plans

We’re considering it

We already have one

13How much risk is your leadership comfortable with?

CautiousStart tight and loosen as we learn

BalancedEasy everyday use, careful with anything risky

EagerWe want staff experimenting and sharing what works

14Who will own this policy?

One named person who answers questions, keeps the tool list and handles problems. Policies without an owner tend to go stale within a year.

Their role

Their name (optional)

Who should staff check with in the “check first” cases?

The policy owner

A small AI working group, one person per department

Their own manager, with the owner as backup

15What role should the board play?

The board approves the policy

The executive director approves it and the board is informed

It’s a staff policy; the board needn’t be involved

16When does it start, and how often should it be reviewed?

AI tools change every few months. A short review cycle keeps the policy honest.

Start date

Every 6 months

Every 12 months

Answers are kept in this browser so you can come back to them.