The big idea: A one-page workplace AI policy names what data never goes into an AI tool and which tools are approved. It says when people need to share that they used AI, who checks the work and who to ask for help. Keeping it to one page makes it easy to read, easy to remember and easy to update.

It's ten minutes before five on a Friday, and you're finally closing your last email. Then a message pops up from someone on the sales team. "Quick question, is it okay if I paste our client notes into a free AI tool?" You stare at the screen, and a small knot forms in your stomach. Nobody ever wrote down an answer.

If that knot feels familiar, you're far from alone. We've watched the same scene play out in companies across every industry. A manager opens a shared drive and finds three different AI experiments already running, with no rules written down anywhere. The tools are moving faster than the guidance, and the real risk lives in that gap.

A one-page workplace AI policy closes that gap. It helps before the gap turns into a rules problem, a data leak or a quiet loss of trust.

We'll give you our answer plainly. The policy belongs on one page, because a page is what people really read, remember and follow. It should name what data never goes into an AI tool and which tools are approved. It should also say when people need to share that they used AI, and who checks the work. That's the whole shape. It clears up confusion without becoming a rulebook nobody opens.

We come at this from two sides of the same table. Coach Shayla spent decades leading people analytics at large global companies. She shared what employees really felt with boards and top leaders. Coach Carlton spent two decades selling software to senior leaders who had to approve tools they barely understood.

Between us, we've seen what happens when a policy is too long to read. We've also seen what happens when there's no policy at all. A long policy brings confusion, and no policy brings quiet risk that nobody manages.

Your people are already using AI, whether or not anyone said yes. KPMG and the University of Melbourne shared a 2025 finding. They found that 44% of U.S. workers use AI at work in ways their employer hasn't approved. That isn't a discipline problem. It's a clarity problem, and clarity is something you can create in one afternoon.

A short policy doesn't hold your team back. It frees them to use AI well, because they finally know where the lines are.

This is part four of our five-part series on AI rollout for managers. Maybe you've been handed the AI job and told to figure it out. This piece can turn that vague worry into a page you can hand out this week.

Key takeaways

  • A one-page policy works because people really read, remember and follow short rules.
  • Name the data that must never go into public AI tools, like personal details, trade secrets or client information under an NDA.
  • List the AI tools people may use, and say clearly when they need to share that they used AI.
  • Ask a person to check all AI-assisted work for accuracy, bias and tone, and name who owns the final result.
  • Choose one person or team, like HR or legal, to answer AI policy questions.
  • Have legal or compliance review the final draft before you share it.

Why a One-Page Workplace AI Policy Is Your Best First Move

Keeping it short isn't a compromise, since short is what makes a policy stick. A team of four or forty can read one page in under two minutes. They can remember it without a refresher and use it the moment a tool opens on their screen. The shorter the policy, the better the odds that people use it.

We've watched long rule books die in shared drives. They get written and approved, and then everyone ignores them, for a simple reason.

A thirty-page policy asks a busy manager to become a rules expert before they can answer a simple question. A one-page policy asks them to glance at five lines and keep working.

The difference isn't the tone. It's how much effort each one asks for.

The story below is an illustration, not a client case. Maya Ellison is an operations manager who saw this happen in real time. Her team was already pasting meeting notes and drafting emails in free AI tools, quietly and without a shared rule. She didn't need a binder.

We use made-up examples like Maya to protect the identities of real employees and businesses.

She needed a single sheet that named what was allowed, what wasn't and who to ask. That sheet became the thing people really used, because it didn't punish them for wanting to move fast.

Maya Ellison holding a single blank-backed sheet in a quiet office corridor
Maya Ellison prepares one clear policy page. Meet Maya

A one-page format makes you decide. It strips out the hedging and the committee language, so you have to name these five things:

  • What data never goes into a public tool
  • Which tools are approved
  • When people need to share that they used AI
  • Who checks the AI work
  • Who to ask when something feels unclear

Five lines for your AI policy

  1. What data never enters a public tool
  2. Which tools are approved
  3. When disclosure is required
  4. Who reviews the AI output
  5. Who to ask when something is unclear
The five policy lines from Maya's story. Meet Maya

That list is the whole policy. Nothing else earns a place on the page, and that's what makes it so strong.

format_quote Worth sharing

When every word has to fight for space, you end up with a page people can hold in their heads. We'd rather a team remember four rules imperfectly than forget forty rules perfectly.

A one-page policy has a second benefit that people rarely mention. It's easy to update. AI tools change every few months, and a short page can be revised in an afternoon without a big committee.

That keeps the policy alive instead of letting it gather dust. What goes into those five lines is where the real work begins.

Writing this page is one of the deliverables inside Workplace AI Academy. It's our self-paced school for managers who've been handed the AI rollout. Each module ends with one real deliverable. One of them is a signed one-page AI policy that covers use, sharing AI use, review and who owns the result. It's one page, not a 40-page legal document.

Data Rules: What Never Goes Into AI Tools at Work

Public AI tools can keep what you type into them, and you don't control what happens next. The data rule that matters most in any one-page workplace AI policy is simple. Personal details, unreleased financials, trade secrets and anything protected by a client NDA never go into a public AI prompt. Text can leave your company's systems and land in a free tool. Then it can be stored, read by others and used to train future models. We can't call that data back, and neither can your legal team.

Coach Shayla spent decades leading people analytics at large global companies. She shared what people really felt with top leaders and boards. That work taught us something uncomfortable. Most data leaks aren't meant to harm anyone, and they usually happen by accident in a busy moment. The same 2025 KPMG study found that 46% of U.S. workers have uploaded sensitive company information to public AI tools.

A manager pastes a performance review into a tool to get help with the wording. A sales lead drops a client list into a prompt to tidy it up.

Nobody meant any harm, but the harm is done the moment the data leaves the building.

So we draw the line before the tool opens. Maya and her team member Jamal agreed that client names, project budgets and employee performance reviews stay out of every AI tool. That talk took ten minutes, and it saved them from the slow trouble of guessing later.

The policy doesn't need to list every possible mistake. It needs to name the kinds of data that cost the most to lose. That means personal details like names, addresses and Social Security numbers. It means private company data like trade secrets and financial results that aren't public yet. It also means confidential client data that sits behind an NDA.

We like one simple test. Before anyone pastes text into an AI tool, they ask one question: would I write this on a public whiteboard? If the answer is no, the data stays out. That one question does more than a ten-page policy, because it turns a rule into a habit.

So the page needs a short list of red lines:

  • Names, addresses, Social Security numbers and other personal details
  • Unreleased financial results, budgets and forecasts
  • Trade secrets, source code and internal strategy documents
  • Client names, contract terms and anything covered by an NDA
  • Employee performance reviews, salary details and discipline records

That list isn't legal advice, and we always encourage managers to have their own counsel review the final wording. Still, these kinds of data stay the same across industries, and they give a team something clear to remember. Next comes the question of which tools are approved. A data rule only works when the tool is chosen with the same care.

We've watched a single pasted paragraph undo months of trust with a client. The policy exists so that never happens on your watch.

Approved Tools and When to Share That You Used AI

A data rule without a tool list is like a fence with no gate. It tells people what not to do, but it gives them no way to get the work done. So the second half of a practical AI policy names the tools your team may use. It also says plainly when a person needs to share that AI was involved. We suggest keeping both lists short enough to fit on the same page as the data rules.

Maya's team chose a simple setup. They listed two approved tools. One was the company's internal AI assistant for anything involving internal documents, and the other was a grammar tool for editing. That was the whole approved list.

Everything else fell outside the line, including the free public AI tools people had been trying on their phones. The policy didn't argue about whether those tools were good or bad.

It simply said they weren't approved, so no one had to guess.

We've watched teams stall on this step because they want to test every tool on the market. You don't need to, since the goal isn't a complete catalog.

The goal is to name the two or three tools that cover most of the work. Then the policy can grow as the team learns. Coach Carlton spent years selling SaaS to senior leaders, and he saw the same pattern again and again. Teams that picked one tool and used it well did better than teams that kept trying out new ones.

When people need to share that they used AI

Sharing rules work best when they're tied to clear moments, not vague ideas. Without them, people stay quiet, and KPMG found in 2025 that 53% of U.S. workers avoid saying when they've used AI. Maya's team wrote three clear moments. First, any outside email drafted with AI gets a short note saying so.

Second, any report that leans on AI content says so in the report. Third, when someone uses AI to make something with their name on it, they check it and own it.

That last point matters, because it links sharing AI use to the human review step that comes next.

  • Outside messages drafted with AI: add a short note
  • Reports or analysis using AI content: say so in the document
  • Work that carries a person's name: check it and own the final version

We also think the policy should keep company tools and personal ones apart. A personal account on a public AI site follows its own terms and privacy rules, not your company's. So the policy should say that company work happens only in tools the company approved, even when a personal tool feels faster. That one sentence stops the quiet slide toward the easy way that weakens so many rollouts.

Before you name tools, it helps to know what work they're for. Our Team Workload Snapshot helps you write down where your team's week goes right now, dated, in tasks and hours. In about two hours, you'll have the starting point your rollout is missing, so your tool list covers real work.

Team Workload Snapshot
Start with your team’s real work

Team Workload Snapshot

See where your team needs support and where AI could save time.

Get the Team Workload Snapshot

Being open about AI isn't about policing people. It's about making sure a client, a director or a coworker can tell a person's judgment from a machine's suggestion. When that line is clear, trust holds.

A Person Checks and Owns Every AI Result

When a team skips the human check, the first AI mistake that reaches a client can sink the whole policy. Every piece of AI-assisted work needs a named person who checks it for accuracy, bias and tone before it leaves the team. That person owns the final result, and the tool never does.

AI can hallucinate, which means it can make up a fact, a source or a number that sounds real but doesn't exist. We've watched a confident sentence about a vendor's pricing slip into a client deck because nobody checked it. More training on the tool won't fix that. A review step will, because it treats every AI draft as a first draft and never a finished one.

So what does the review look for? It checks three things, in this order:

  • Check that it's accurate. Does every claim match a source the person can name? If a number appears, where did it come from?
  • Check it for any bias. Does the wording assume a gender, a background or a point of view that doesn't fit the reader or the moment?
  • Check that the tone fits. Does it sound like the person whose name is on it, or like a machine pretending to be sure?

Dana Kim raised the question of ownership while Maya's team drafted their policy. Her point was simple. If three people touch an AI draft and nobody owns it, mistakes become everyone's problem and no one's fix. So the policy names the person who ran the tool as the one responsible. That person signs off before anything moves forward.

Helen Brandt, Maya's director, drew the same line for anything a client will see. A report first drafted by AI still gets a careful review and a sign-off from the person responsible. That isn't extra red tape. It's the difference between a tool that supports good judgment and a tool that quietly replaces it.

We also know some review questions will come up that one person can't answer alone. When the policy names a go-to person for those moments, the review step keeps moving. Doubts go to that person instead of slipping into the work.

Maya Ellison reviewing a blank-backed sheet at a writing desk in warm afternoon light
Maya Ellison reviews a draft before she shares it. Meet Maya
format_quote Worth sharing

People handle nuance, context and ethics, while AI handles speed and patterns. The policy makes that split clear, and naming an owner makes it stick.

Who to Ask: Getting Support for Your Workplace AI Policy

A policy only works if people know where to take their questions. Name one person or one team as the first stop for anything about AI, and put that name on the page itself. HR is the most common choice, but legal, compliance or a chosen AI lead can work as well. What matters is that nobody has to guess.

We've watched teams stall because the policy existed but the path to a real person didn't. Someone hits a gray area, asks three coworkers and gets three different answers. Then they quietly stop using the tool, and that silence is worse than the question. A named contact turns that doubt into a ten-second answer.

Maya decided that her HR team would be the first contact for AI policy questions. She wrote it as one line near the bottom of the page: "Questions about this policy go to HR." She didn't need a committee or a shared inbox that nobody checks. One clear door was enough.

Before anyone signs off, the final draft needs a legal or compliance review, and it's far more than a formality. Job laws, data privacy rules and industry rules all touch AI use. A missed clause can cost more than a slow rollout. We'd say the same thing to any manager: your policy is a draft until your own counsel has read it.

Maya's director, Helen Brandt, made sure the company's legal team reviewed the final draft. That one step changed the page from a well-meaning internal note into something the company could stand behind. It also helped Maya when someone pushed back on a rule later. She could point to the review and say the policy had been checked.

Maya Ellison holding a closed unmarked folder beside an office door in evening light
Maya Ellison gets her draft ready for legal review. Meet Maya

Legal review also catches things a busy manager can miss. A phrase that sounds clear to a team may mean something special in a contract. Counsel can tighten the wording without adding length. They can also flag rules that clash with other company policies. That's their job, and it's worth the wait.

Once the policy is live, updates should come through the same person. When a tool changes its terms or a new data rule appears, the named contact should be the one who shares it. A policy that never changes starts to feel like a museum piece, while a policy with a clear owner stays alive.

If you're that named contact and you'd like to think it through with someone first, book a Discovery call with us. It's a 30-minute conversation with Coach Shayla and Coach Carlton about where your rollout feels stuck and what to do next. It starts with a small deposit that is credited toward your next call or the 5-Day Challenge.

format_quote Worth sharing

We'd rather see a one-page policy with a named contact and a legal review than a ten-page policy with neither. The structure is what makes the rules hold.

Write the Page, and Let Your Team Breathe

A one-page workplace AI policy works because it takes away the guessing. Your team stops wondering whether a tool is allowed or what data can leave the building. They know who checks the work before a client sees it. Clarity creates freedom, and one page gives everyone the same answer at the same time.

We've watched good AI rollouts stall for one reason: nobody wrote down the rules before the tools arrived. The policy isn't paperwork. It's the guardrail that lets your team move faster without fear.

Picture next Friday at ten minutes before five. The same message pops up, and this time you smile, paste one line from the policy and close your laptop. Your weekend starts with a calm mind instead of a knot in your stomach.

These are the pieces that belong on that page:

  • Write down the data rules. Name what never goes into an AI tool, like customer details, employee records, unreleased financials and anything confidential.
  • List the tools you approve. Name the tools your team may use, and say clearly that anything else needs a talk first.
  • Say when to share AI use. Explain when someone should mention that they used AI, especially in client work or choices that affect people.
  • Ask a person to review it. Have someone read, check and own every AI result before it goes anywhere.
  • Make one person the go-to. Name who to ask when someone feels unsure, and make that person easy to reach.

Before you finalize anything, send the draft to legal or compliance. That review isn't a delay. It's the step that turns your good intentions into something your company can stand behind.

Then start with what you already know. Before you write rules for AI, see where your team's week really goes. It takes one afternoon and one dated page, with no tool named. That starting point shows you which tasks really need AI, so your policy covers real work instead of guesses.

Write the page, date it and share it. Then your team can get back to the work that matters, with clear lines to keep everyone safe.

Where to start this week

Begin with where things stand today. The Team Workload Snapshot takes one afternoon and gives you one dated page of where the week goes. When you're ready to run the whole rollout, join Workplace AI Academy. It takes you from that page to a signed policy and a return summary for your leaders. You'll work through it at your own pace. And if you'd rather talk it through first, book a Discovery call with Coach Shayla and Coach Carlton. It's 30 minutes with both of us. It starts with a small deposit that is credited toward your next call or the 5-Day Challenge.

Characters in this story
Maya EllisonMaya EllisonAn operations managerMeet Maya →

Frequently asked questions

Why do we need a one-page workplace AI policy now?

A short policy sets clear rules for using AI tools, so company data stays safe and everyone is treated fairly. It helps teams use AI with confidence and avoid mistakes and legal trouble. That clarity matters more each month as AI tools show up in daily work.

What are the biggest risks if we don't have an AI policy?

Without a policy, people may paste private information into public AI tools and cause a data leak. Unchecked AI work can lead to biased choices, and each team may use AI in a different way. All of that can bring legal trouble, harm your reputation and wear down trust.

Can a one-page policy really cover everything we need?

Yes, it can. A one-page policy covers what matters most. It names what data stays out, which tools are approved and when to share AI use. It also names who checks the work and who owns it. Because it's short and clear, people are more likely to read and follow it than a long document.

Who should help draft our team's AI policy?

Bring in HR, legal or compliance, IT and the managers who lead teams using AI. Including team members helps, too, the way Maya's team worked with Jamal and Dana. Their input keeps the policy practical, so more people follow it.

What should I do before I write the policy?

Write down where your team's time goes today. The Team Workload Snapshot walks you through it in nine short sections, about two hours in all. It comes as a Notion page you copy, plus a printable PDF.

Where can I learn to write and roll out the policy step by step?

Workplace AI Academy is our self-paced school for managers, HR leads and operations leaders. You build a team data rules card, a signed one-page AI policy and the rest of the rollout. Each module ends with one deliverable. It isn't legal advice, so your own counsel still reviews the policy before you share it.

Can someone look at my draft with me?

Yes, we'd be glad to. Book a Discovery call and talk it through with both of us. It starts with a small deposit that is credited toward your next call or the 5-Day Challenge.

Sources

  1. KPMG and University of Melbourne, Trust, Attitudes and Use of AI: A Global Study 2025, U.S. findings release, April 29, 2025
Team Workload Snapshot
See where your team needs support

Team Workload Snapshot

helps you see where your team needs support and where AI could save time.

Get the Team Workload Snapshot
Invisible Limits Finder™
For the beliefs behind quiet resistance

Invisible Limits Finder™

helps you find the hidden rule holding you back, so you can move forward.

Get the Invisible Limits Finder™
Workplace AI Academy
For your workplace rollout

Workplace AI Academy

Get your team using AI

Explore the school
See everything in the shop