If you haven’t approved an AI tool for your business, your staff are probably using one anyway. In Microsoft and LinkedIn’s 2024 Work Trend Index, three out of four knowledge workers said they use AI at work, and 78% of those users said they bring their own tools rather than waiting for their employer.
The usual reaction is to ban it. That rarely works, and it hides the risk instead of removing it. Here’s a practical sequence for a 10–100 person business: find out what’s already in use, pick one approved tool, and write a one-page rule your team will actually follow.
Myth: a ban keeps your data safe
A ban stops people from mentioning AI. It doesn’t stop them from using it. Someone who saves an hour a day drafting emails won’t give that up because of a memo. The use just moves to personal phones and home laptops, where you can’t see it and can’t set any rules.
The reality: you can’t control what you’ve driven out of sight. The businesses that handle this well treat AI the way they treat email. It’s allowed, it runs on a company account, and there are clear rules about what goes in it.
The real risk is what gets pasted in
The danger isn’t that an employee asks a chatbot to tidy up a paragraph. The danger is what’s in that paragraph. Customer names and account numbers, a spreadsheet of payroll figures, a contract with pricing, a patient or student record. On free consumer accounts, providers may use what you type to improve their models unless the user finds the setting and turns it off, and your business has no admin control, no record, and no way to delete it when that person leaves.
If your business is covered by HIPAA, holds card data, or has confidentiality clauses in client contracts, one careless paste can breach obligations you’ve already signed.
1. Find out what people already use
Start with a short, no-blame question to the whole team: which AI tools do you use for work, and what do you use them for? Make it clear nobody is in trouble, and you’ll get honest answers about which jobs AI is doing for you, such as emails, summaries, or spreadsheet formulas. Then have your IT provider check firewall logs and the apps users have connected to their Microsoft 365 or Google accounts, which show tools nobody mentioned.
2. Pick one approved tool with a business account
The key difference is between consumer and business versions. OpenAI says data from its business plans, ChatGPT Business and Enterprise, isn’t used to train its models by default. Microsoft says that when staff sign in to Copilot Chat with a work account, it comes with enterprise data protection and prompts aren’t used to train its foundation models. Google makes a similar commitment for Gemini in its Workspace business plans.
What good looks like: one approved tool, everyone signed in with their company account, and admin controls you can manage. If you already pay for Microsoft 365, Copilot Chat with work sign-in is the obvious place to start. Confirm current terms on the vendor’s own site before you commit, because they change often.
3. Write the one-page rule
Your policy should fit on one page and answer three questions: which tool is approved, what may never go into any AI tool, and who to ask when you’re unsure. The “never” list does most of the work:
- Passwords, MFA codes, or anything that unlocks an account
- Customer or patient personal details: names with account numbers, Social Security numbers, health information
- Payment card and bank details
- Payroll, HR, and personnel records
- Anything covered by a client contract or NDA
Add one more line: a person checks anything AI produces before it goes to a customer. AI tools state wrong facts with total confidence, and your name is on what gets sent.
4. Teach people to spot fake AI apps
Criminals know everyone is searching for AI tools. In May 2025, Google’s Mandiant team reported a campaign that used thousands of social media ads on Facebook and LinkedIn to send people to more than 30 fake “AI video generator” sites. The download installed malware that stole saved passwords, browser cookies, and credit card details.
The rule is simple. Only get AI tools from the vendor’s own website or the official app store, never from an ad or a link someone sent you, and ask before installing anything, browser extensions included. If a free tool asks you to download a program in order to see your results, close it.
When a ban actually makes sense
Some businesses should restrict AI tightly, at least for now. If your work is mostly regulated records and you haven’t confirmed a tool’s terms meet your obligations, such as a signed HIPAA business associate agreement, “not yet” is a reasonable answer. Even then, say so in writing and give people a date when you’ll revisit it, rather than leaving it unspoken.
The honest summary
Your team uses AI because it saves time, and that won’t stop. The choice is whether it happens on accounts you control, under rules people understand, or in the dark on free personal logins. One approved tool, a one-page rule, and a habit of downloading only from official sources will cover most of the risk for a business your size.
Not sure what your team is already using? Book a free 30-minute consult and we’ll help you find out what’s in use, choose an approved tool that fits the systems you already pay for, and draft your one-page rule — and tell you honestly whether you need anything more than that.