AI Use Policy
| Approved by | Board of Directors |
|---|---|
| Effective | Date approved |
| Next review | 12 months after approval |
| Policy owner | Executive Director |
| Applies in | Ontario |
1. Why we have this policy
AI tools like ChatGPT and Microsoft Copilot can save our team real time. They also make it easy to send private information about the people we serve to companies outside our organization. This policy lets us use AI where it helps and keeps that information safe.
2. Who it covers
Everyone who works with or for our organization: staff, volunteers, students on placement, board members and contractors. It covers any AI tool, on a work or personal device, whenever you are doing work for our organization.
3. Tools we use
- You may use ChatGPT and Microsoft Copilot for work, following the rules in this policy.
- Ask the Executive Director before you start using any other AI tool, browser extension, plugin or meeting note taker for work. Many of them can read everything on your screen or in your inbox.
- Where a tool lets you stop your chats being used to train its AI, turn that off. In ChatGPT it is under Settings, Data controls, "Improve the model for everyone".
4. Green, yellow, red
Green: go ahead
- Drafting newsletters, social media posts, event invitations and job postings
- Brainstorming, outlines, and improving your own writing when it has no personal information in it
- Summarizing public reports, research and funder guidelines
- Grant applications and reports, as long as no client stories or details go in
- Spreadsheet formulas, templates and checklists
Yellow: ask the Executive Director first
- Anything with donor, volunteer or staff personal information in it
- Recording or transcribing a meeting with an AI note taker. Everyone in the meeting must agree first
- Publishing anything mostly written by AI under our organization's name
- Connecting an AI tool to our organization's email, files or databases
Red: never
- Client information in any AI tool on a personal or free account. That means names, phone numbers, emails, addresses, dates of birth, SIN, health card and file numbers, photos and case notes
- Letting AI make a decision about a person: eligibility for services, risk or safety assessments, hiring or discipline. AI can help with a draft. A person decides
- AI note takers in client sessions, case conferences or HR meetings
- Making images, video or voice recordings of a real person with AI without their permission, or passing off AI images as real photos of our clients or programs
5. Using AI with client work
AI can help with client work, like tidying a case note or drafting a referral letter. Before anything about a client goes into an AI tool:
- Take out every detail that could identify the person: names (including first names and nicknames), contact details, ID numbers, dates of birth, addresses, and any description that only fits one person, like "the only Tigrinya speaking mother of triplets in our program".
- Read it before you paste it. A free tool like Safe Paste can find and hide most details for you, but a person always does the final check.
- Put the real details back only in our own systems, never in the AI tool.
6. A person checks everything
- AI can be confidently wrong. Check every fact, number, name, date and quote before you use what it wrote.
- Never pass on AI answers as legal, medical, immigration or financial advice to a client.
- Watch for bias and stereotypes about the people we serve, and fix them.
- You are responsible for anything you send, post or file, no matter who or what drafted it.
7. Being open about AI
- Tell people when AI did a large part of something they would want to know about, like a report, a published article or an image.
- If AI is ever part of a service our clients receive, we tell them in plain language and they can always talk to a person instead.
- If a funder asks whether AI was used in an application or report, answer honestly.
- If we have 25 or more employees and use AI to screen, assess or select applicants for a publicly advertised job, the job posting says so, as Ontario's Employment Standards Act requires.
8. Accounts and security
- Use a strong, unique password and two step login on every AI account you use for work.
- Do not keep chats with work information in them on a personal AI account. Delete them, and delete any that remain when you leave.
- Report a lost device or a hacked account that had AI chats with work information to the Executive Director right away.
9. Laws we follow
No single law covers every nonprofit. These are the ones most likely to apply to us in Ontario. The Executive Director confirms which ones apply to our work.
- Personal Information Protection and Electronic Documents Act (PIPEDA). It covers us only for anything we do that counts as commercial activity, like selling, bartering or leasing donor or membership lists. Most of our charitable work is not covered by it, and we protect personal information to that standard anyway.
- Employment Standards Act, 2000, section 8.4 (ESA). If we have 25 or more employees, our public job postings must say when AI is used to screen, assess or select applicants (see section 7).
- Our duty of confidentiality to the people we serve, and the professional standards of any regulated staff, such as social workers.
Privacy regulator for the laws above: the Office of the Privacy Commissioner of Canada.
10. If something goes wrong
- If you put information into an AI tool that should not have gone in, tell the Executive Director the same day. Reporting fast matters more than anything else, and you will not be punished for reporting honestly.
- The Executive Director finds out what happened, has it deleted where possible, decides whether the law requires us to tell the people affected or a privacy regulator, and keeps a record of every incident.
11. Questions and review
- Questions about this policy go to the Executive Director.
- We review this policy every 12 months, and sooner if the law or our tools change.
- Everyone it covers reads it and signs below when they start at our organization, and again after each update.
I have read this AI Use Policy and I will follow it.
Name: Signature: Date:
Made with the free AI policy builder from Safe AI for Nonprofits. It is a template to adapt, not legal advice.