AI tools have arrived in most workplaces faster than anyone wrote rules for them. Someone on your team is probably already using one to draft an email or summarize a long report. Most of the time that is a good thing. The trouble starts when nobody has said out loud what is fine and what is not.
Why each rule comes with a reason
These five rules come from Module 1 of our AI training at TeamLogic IT (watch Module 1), and they fit on a single page (the five golden rules handout). We explain each one, because a rule someone understands survives contact with a busy Tuesday. A rule they merely obey does not. So below, each rule appears exactly as it reads on the handout, followed by the reason behind it and what it looks like on an ordinary workday.
1. Use the company account
Use the company account. Work goes through the tools and accounts the company provides. Personal accounts are for personal life.
When you sign in with your work account, your company can see which tools are in use, control who has access and put protections in place behind the scenes. A personal account has none of that. If you leave the company, anything you pasted in leaves with you.
Picture a sales coordinator who polishes a client proposal in a free chatbot on her phone because it was already open. Nothing bad happens that day. But the proposal now sits in an account the business doesn't know about and can't manage. The company account takes the same few seconds and keeps the work where it belongs.
2. Know what never goes in
Know what never goes in. Customer data, pricing, contracts, employee records, passwords, and anything under a confidentiality obligation — unless the tool is approved for it.
Typing something into an AI tool is a lot like sending it to an outside company. Depending on the tool and its settings, what you paste may be stored or kept longer than you expect. For most of your work, that is fine. The items on this list are different. Many are covered by contracts, privacy laws or promises your company has made to its customers.
The closing words matter too. Some tools are approved for sensitive work, and there the rule bends on purpose because someone has checked first.
Say you want help writing a polite late-payment reminder. You could paste in the customer's name, account number and balance. Or you could ask for a template with blanks and fill them in yourself. The second way gets you the same email without the customer's details leaving your hands.
3. Review before it leaves you
Review before it leaves you. Anything sent, published, or acted on gets your eyes first. You own the output.
AI writes with the same confidence whether it is right or wrong. It can misread your request, pick the wrong tone or add something you never asked for. That is harmless on your screen and costly in a customer's inbox.
"You own the output" is the heart of this rule. If an email goes out under your name, the customer will not care which tool wrote the first draft. It is your email.
Imagine a service manager asking AI to summarize a long email thread for the team. The summary reads well, but it drops the one detail that changes the plan: the customer moved the install date. A two-minute read before sending would have caught it.
4. Verify facts that matter
Verify facts that matter. Numbers, names, dates, citations, legal or technical claims — check them against a real source.
AI tools predict what a good answer should look like. Usually that lines up with the truth. Sometimes it produces a figure, a date or a reference that sounds right and simply isn't. The tool will not flag its own mistake, so the check has to come from you.
This doesn't mean fact-checking every word. It means checking the facts that would cause harm if they were wrong.
If AI helps you write a quote and lists a product's warranty as three years, confirm it on the manufacturer's page before the quote goes out. Checking takes a few minutes. Correcting a wrong number after a customer has relied on it takes much longer.
5. Ask when unsure
Ask when unsure. New tool? Odd request? Not sure if data is safe to use? Ask your AI point of contact before, not after.
New AI tools keep appearing, and real work throws up situations nobody planned for. This rule is the safety net under the other four. It only works if asking feels normal, so learn who your AI point of contact is before you need them.
"Before, not after" is the important part. A question asked first costs a minute. The same question asked after the data is in the wrong place turns a quick answer into a cleanup job.
Suppose a new browser add-on promises to summarize every page and email you open. A short message to your point of contact asking whether it is okay to install is the right move. That is much easier than explaining later why the add-on could read your inbox.
Where these rules go next
Taken together, these five rules are your company's acceptable-use policy in draft form. They cover which tools to use, what stays out, who checks the work and where to go with questions. Module 5 is where they become the written one.
Until then, three small steps put the rules to work:
- Check that you are signed in to your AI tools with your work account.
- Find out who your AI point of contact is.
- Keep the handout (five golden rules handout) somewhere you will see it.
If you would like to talk through how these rules apply to your team, let us know. We are happy to help.
Want help putting this to work?
We help small businesses in North Alabama, Middle Tennessee and Knoxville use AI safely and get more from the technology they already have.