In the past two years or so, I’ve had countless conversations with shops regarding AI. From my perspective, getting into AI use looks just like a kid getting into a pool. First, a toe dips in. Then, it’s a running and jumping cannonball from the back deck.
You may have started last year by having AI write an email or a Facebook post. Now, here we are, building quotes, curating your email, designing artwork, and analyzing customer information. And an AI Agent that does work for you without you stepping in or making a decision is right around the corner.
So now is a good time to set some guardrails for your business regarding how employees use AI. It’s not hard. But it does require some thought.
First, let’s determine the rules for how your team uses AI at work. Which tools are ok? On what type of task? On what particular sets of data or information? And let’s clearly define who is responsible for the result.
Right now, I’ll bet there is not a single word in your employee handbook about AI.
What you want to create is a single source of truth about AI that outlines how your team can do great work without guessing or fear.
And, because I know you want it, I’ve added a complete copy-and-paste version of an AI policy you can use today at the end of the article. Just modify it to make it your own, with how you want AI used. You can also take this and use it as a starter prompt in AI. It will work great.
Let’s walk through what belongs in your Employee Handbook AI Section.
A Quick Note Before We Start
First, the obligatory legal disclaimer. I’m a decorated apparel industry business consultant. Not an attorney. Everything contained in this article is a starting point to help you think, not legal advice.
Fact: AI law is changing fast. The rules vary from state to state and from country to country. Before you finalize any word of the new section of your employee handbook on AI, you might want to run it past your business attorney first. The intent of the copy-and-paste block at the end of the article is to give you something to build on, not a finished legal document.
Good? Awesome. Let’s go.
AI Is Just A Tool
I know to some people AI is the boogeyman, the devil, and evil all wrapped up in one. To me, it is a tool. Just like a shovel is a tool. Photoshop is a tool. And a can opener is a tool.
Tools are made to get a result. And much like any other tool in your building, they work great in trained hands, but can be a complete disaster in careless ones.
I want to be clear on how I stand on the ethics of AI, because there is some chatter about this. AI, whether it is a natural language tool like Claude or ChatGPT, or an image generator like Midjourney or Nano Banana, is trained on vast amounts of data and information. Whether you like it or not, that genie isn’t going back into the bottle.
Long before AI, people have learned from and been inspired by the work and talents of others. Your employee handbook policy isn’t going to turn a corner-cutter into a saint.
A policy is a decision made in advance.
Your AI policy, as expressed in your employee handbook, tells your team how your shop expects all AI tools to be used. The mission of the policy is to define what is ok, and what is not. If you don’t want someone else’s intellectual property to be used with AI, you have to say so. If you want to keep your customer data out of public AI databases, you have to say so.
This is new stuff. You can’t expect “common sense” when someone is using it, so you have to train your crew on what is allowed and what is not. It is as simple as that.
What is right? What is wrong? What happens if you are unsure? It’s all in there.
Don’t Write A Policy About The Tools
Here’s a big one. Do not build your policy around the names of today’s tools or what you are currently using.
AI is changing quickly. Your favorite tool you are using today might not cut the mustard 6 months from now. If you name the tools in the policy, it will be obsolete the second you switch tools or start testing a new one. And these days, a new one is always surfacing.
Therefore, write your policy with usage guardrails instead. Guardrails don’t expire. A good AI policy answers a few durable questions no matter what you are using:
- Company accounts only. Imagine the problems down the line when that app, that design, or that content was created with an employee's personal account? The company should pay for the tools to be used professionally. If the company won’t, then they can’t be used.
- Define Usage. What is allowed, and what is the defined outcome?
- Data. What can go in, and what can’t? Be hyper-specific here.
- Accountability. Who checks the result for accuracy? Who owns the result? Who is in charge? What happens when there are mistakes?
- Output. When something is finished, how are you using the output or outcome?
Please notice that none of those 5 points care which AI tool is being used. That’s the point.
And one more thing. Don’t ban AI. Maybe you are on the side of the coin that isn’t a fan. I get it. A ban feels safe.
Let me tell you, though: a ban won’t stop AI use; it just pushes it into the shadows where you have zero visibility or control. You are far better off giving your employees approved tools, training, and clear rules than pretending AI tools don’t exist.
Also, while we are discussing this, if you are the business owner creating this policy, I think you need to actually use these tools yourself. You can’t set smart rules for something you have never touched or don’t understand. Spend some time on your own to understand what is involved, how the AI tool works, and to research best practices before you set guardrails for how your team’s usage should be controlled.
Lastly, as this is a policy, who owns this once it is written? Who is going to enforce it? To me, it should be your executive leadership team, HR, and your department heads, the same as every other rule you have in your business. This isn’t an IT project. It is a management responsibility. Give it a real owner, and have them review it on a scheduled basis. This stuff moves fast.
Data: Garbage In, Confident Garbage Out
I’ll bet right now if you typed in “how to analyze data using AI,” you would get an endless list of articles, videos, and a ton of talking heads telling you the best way to do it. I’m certainly not going to add to that pile.
But what I am going to say is that if your data is a mess, AI isn’t going to clean it up before it does its magic. It will hand over a clean, confident, professional-looking answer. And it can be 100% wrong. That’s the trap.
Your employee handbook policy needs to address this. The single most valuable habit you can build into your policy is to verify against the source. AI is excellent at pulling an answer out of a messy pile. However, it is terrible at telling you the answer is wrong. A human needs to be in the loop, especially before you are acting on any number or information gleaned from the data. You have to check.
So before you drop your customer list, pricing, contracts, financial records, employee information, or any other sensitive information into an AI account, be aware that this information can leave your control.
Being careful about how we use AI is the reason why you need a policy. You need to spell out clearly what never goes into an AI tool. Information such as customer or personal information, pricing, financials, contracts, anything under an NDA, and of course, your own proprietary files.
Be clear, state it plainly, so your team will follow it.
Artwork Is Where This Gets Real For Our Industry
Unless you are an apparel distributor like S&S Activewear, just about every order you ship has a decoration on it. Screen printing, digital printing, embroidery, heat-applied, sublimation, or other methods all have one thing in common…there is usually an art file. This is where an AI image might actually touch your actual product. This section might matter most for you, and I have a slightly different take than most of the advice you’ll read.
To me, the discipline is on the input, not the output. Here’s what I mean.
I’m a designer. I started off as an art director in the industry back in 1993. Right before then, I was working on my master's degree in architecture and had a small t-shirt business that catered to fraternities and sororities. How I got sucked into this business was the invention of the Apple Macintosh computer and Photoshop.
Before then, everything was created by hand. It’s hard to imagine now, but one t-shirt design, especially a multi-color one, could take several days and 15 or 20 hours of work. Then I tried out working on a Mac. Just simple things, but I quickly realized that all of that work by hand could now be created with less effort using modern tools. 20 hours of work was now 2 or 3. You still needed to know what you want, and a great starting point was a loosely scribbled thumbnail.
Now, with AI, that 2 or 3 hours is about 20 minutes. What the customer pays you hasn’t changed, but the cost of doing the art most certainly has.
Before AI, if you wanted to copy the style of Norman Rockwell, Frank Frazetta, or Ralph Steadman, you could, but it was difficult. Now, with AI, you just prompt “In the style of Norman Rockwell,” and it is easy. As a designer, I want to look like 100% me all of the time. The tools make it easier to emulate, but that doesn’t mean it is correct.
And frankly, it is unnecessary. Using AI, you can create images in your style that fit your thumbnail and complete your vision, without borrowing anything from Rockwell, Frazetta, or Steadman. AI isn’t going to change how ethical someone is when they use the tool. Nor should it. I don’t know if you remember, but when Photoshop came out, there was backlash from photographers who complained that the “image was Photoshopped.” Now? Using Photoshop is expected, because it makes images look better.
To me, AI is a professional tool just like Photoshop. Or a kitchen knife. Or anything else. Don’t use it correctly? You might get in trouble. Set some guidelines and rules for your creative team. Don’t use anyone else’s intellectual property. Start with your own idea. Upload and use your own source material. Basically, just like before AI.
Art Workflow and Copyright
Currently, you can’t copyright AI art. You also can’t copyright written text or code either if it was created by AI.
In the United States, the Copyright Office has been clear that work created purely by typing in a prompt is not yours to copyright because a human didn’t create the expressive part. However, the same office has been just as clear that when a human selects, arranges, and modifies AI-generated pieces into something new, that human contribution is protected.
So if you have a thumbnail sketch, create a bunch of different elements as parts, edit them, assemble them, and direct the entire project yourself, your work is a lot more yours than “AI made my art” would suggest.
But if you bang out a prompt, take that result, and throw it on a shirt? No dice. Especially if you use someone else’s name (“In the style of Norman Rockwell,” for example), intellectual property, uploaded images, or direct ideas. You know. Just like before AI existed.
Now, some real talk, as some folks are genuinely worried about the copyright question. Answer me this: how much do you think it costs to actually challenge someone in court if they start using your design? That’s what you are worried about, right? Let’s say a competitor steals your idea and throws it up on a website somewhere. It is your exact artwork. Even if you didn’t use AI to create that art, it’s going to take you about $200,000 in legal fees to challenge that through the process. So for a shop your size, is that even a reality? Probably not. Just move on.
Your AI policy should define how the art is created, how it is used and assembled, and what the creation rules are. This is the same policy you should have written before AI even existed.
Own It, or Don’t Use It
Here’s a simple rule that ties this whole thing together. The person using the tool owns the result. Not “the AI.”
If an employee uses AI to create a quote and the final number is wrong, that’s not the AI’s mistake. It’s the employee’s. The same as if they fat-fingered it with a calculator.
Remember, AI can make things up. It can state the wrong information with total confidence. Your policy has to make accountability obvious. Whoever uses an AI tool is responsible for checking the work and standing behind it. A human reviews every AI-assisted outcome, especially ones going to customers. Every time.
And this is the part I feel strongest about. AI usage can’t be a secret.
If you are a professional, using a professional tool, in a professional way, own the result. Don’t be embarrassed or hide it. AI tools are legitimately being used by businesses all across the globe. You have nothing to apologize for or be embarrassed about.
However, if you are uncomfortable with telling your customer you used AI on their project, then that discomfort is telling you something. Can you stand behind your professional use of AI? If you can’t, you shouldn’t be using it.
Own it, or don’t use it.
Also, this is how you empower your team. When people know the rules, have been trained to follow them, and are trusted to use the tools, knowing they won’t be punished for using them well, they can do their best work.
Your policy is not erecting a fence to keep them out. You are giving them a clear field to run on, making their work easier.
Copy, Paste, and Make the Employee Handbook AI Policy Yours
Below is your starting point. The idea is for you to drop this into your next employee handbook update and adjust it to fit your business. I can’t possibly know what you do, so customize it to fit.
Or, upload your current employee handbook into AI, and take what I’ve written below and paste it in as well and say something like “Rewrite this AI policy to fit my business. Be clear, and match our current employee handbook’s tone. Ask me any questions you may have about how we use AI, any guardrails we want to adopt, or anything else you need clarity on before writing the first draft.”
Either way, make sure your attorney reviews it before it goes live with your staff.