November 17, 2026

Here's a policy I've seen. It says the only AI tool staff can use is Microsoft Copilot. Period.
The organization has a Microsoft 365 nonprofit subscription, and Enterprise Data Protection sounds reassuring on paper. But the subscription is the free Business Basic tier, which lacks the advanced protections the policy assumes. And over the years, plenty of SharePoint and OneDrive folders have been set to "Everyone in the organization."
That means Copilot can surface sensitive files to people who should never see them. And it has. Staff members have told me what they found with simple searches.
Nobody follows the policy. I hear staff members say all sorts of things as they reach for their tool of choice:
"We're a Microsoft shop, sure. But our meeting notes come from Google Gemini in Google Meet. We’ve been doing this for years."
“Supposedly the policy bans my ChatGPT agent in a business account. I’ve spent months training it. It gets the work done for me.”
“The policy has been revised eighteen times in three years. It’s sixteen pages long, and only a lawyer can read it.”
Another organization runs some departments on Microsoft and others on Google. There’s no standard. Each department picked its own tools to solve its own problems. As soon as leadership suggests standardizing, the arguments start. So does the trench-digging.
And one more. This policy had a solid security section. It forbade storing sensitive data outside approved systems. But the staff member who pulled together the impact reports couldn't do his job without downloading donor and program data to his personal laptop and running it through his own reporting tool. The enterprise data solution that would have met the policy's requirements was never funded. The reports still had to go out. So the laptop won.
Every one of these is a policy problem. None of them is really a writing problem.
The Gap Between the Organization You Want and the One You Have
A tech policy that staff won't follow almost always describes the organization leadership wishes it had, not the one doing the work. It lists the approved tools, not the tools people actually rely on. It bans the workaround without offering a replacement. It assumes a budget that doesn't exist.
I've been in the room for a lot of these conversations. I've flagged the Gemini dependency. I've pointed out the open folders. I've asked what happens to the impact reports. I rarely hold the pen, and I'm not supposed to. The policy has to belong to the organization, or nobody will follow it.
But I can tell you what I've seen work, and what I'd do differently if I were starting over with a nonprofit.
Here’s What I Would Do
- Find the tools people actually use. And expect to keep finding the tools they use, every six months or so. Ask the staff, plainly and without blame: "What do you use to get your work done, including the things you'd never mention to IT?" You'll hear about the Gemini notes, the personal ChatGPT account, and the laptop spreadsheet.
- Say what's allowed, not just what's banned. A ban with no alternative doesn't stop the behavior. It just moves it out of sight. For every restricted tool, name the approved option or say clearly that none exists yet. Then the policy is a map, not a wall.
- Keep it to one page. Write the principles in plain language. Add a short list of approved tools and the types of data each one can hold. Put the legal detail in an appendix for the lawyers, one nobody has to read to get their work done.
- Pair every "no" with a funded "yes." This is the hardest step, and the most important. If the policy forbids downloading data to personal laptops, the organization has to fund a secure place to do that work. If it can't, the policy should say so. Name the gap, name the project, and name who owns the decision.
- Give the policy an owner and a review date. One person should be responsible for keeping it current, and it should be reviewed on a set schedule. That's how you avoid eighteen revisions in three years. Small, scheduled updates beat big rewrites that nobody reads.
What Success Looks Like
A good policy won't feel exciting. It will feel obvious. Staff will read it once and know what to do.
You don't need a perfect policy to start. You need one that matches reality and gives people a safe way to do their work.
If you're not sure where your organization's policy stands, start with the inventory. Ask people what they actually use. That conversation alone usually tells you more than the document does.
If you want a second set of eyes on your current tech policy, or on the gaps between what it says and what your team does, book a free 30-minute consultation. You can also start with our Data Audit Toolkit, which helps teams see where their information enters their systems and where it breaks down.


