"Who owns AI in a company?" In most companies with ten to two hundred people, the honest answer today is nobody, or everybody, which comes to the same thing. Staff sign up for tools on their own, a vendor switches a feature on in an update, and when a lender, an insurer or a buyer asks who is responsible, the owner looks around the room. The practical answer is simple: one named person owns the day-to-day, with a written scope, and the owner keeps three decisions that never move. This article is for the owner who wants that settled in a week, without hiring a new executive. Whether the company needs a written AI strategy at all, and the five decisions that make one, is in do I need an AI strategy.
I was COO and then President of a public company that closed more than forty acquisitions, and I ran the integration on most of them. I am a lawyer by training. Now I write fixed-fee AI plans for CEOs of private-equity-backed and owner-led mid-market companies through 013 Labs. This is not legal advice. Where a decision touches employment, contracts or data protection, ask your own adviser.
Why "nobody" is the default
AI arrived in most companies through the side door. Someone tried a free tool to write an email; someone else used one to tidy a spreadsheet; a software provider added an assistant to a product you already pay for. Nobody decided any of it, so nobody owns it. That is not a crisis, but it means nobody can answer three questions an outsider will eventually ask: what tools are in use, what data goes into them, and who checks the output.
The one named person
Usually someone you already have. In most companies there is a person others already go to when a new tool shows up. They are curious, careful, and already using these tools for real work. That person is usually a better owner than a new hire. Our free guide, spot your AI champion, sets out how to find them.
A written scope, one page. What they own: keeping the list of tools in use, the data rule, approving new tools, answering staff questions, and reporting to you once a month. What they do not own: decisions that commit the company.
Time, not a title. A few hours a week, protected, is enough to start. If the work grows, the scope grows with it. Should I hire someone for AI or train my staff covers when a hire actually makes sense.
The three decisions the owner keeps
First, what never goes into an outside tool. Customer lists, pricing, payroll, bank details, anything confidential. One sentence everyone can repeat. The named person enforces it; the owner decides it.
Second, what a tool never does alone. Prices and quotes, contracts, payments, hiring, and anything signed in the company's name. A person always does the last step.
Third, what gets built or bought. Spending money on a new system, or giving a tool access to company data, is the owner's call, on the named person's recommendation.
What it looks like on paper
One page with: the named person and their scope; the data sentence; the never-alone list; the tools in use today and who approved each one; and the date of the next monthly review. That page is also most of your AI policy. Do I need an AI policy for my small business shows how to turn it into the policy staff sign.
When outsiders ask
A bank, an insurer, a buyer or a board increasingly asks who is responsible for AI. "Our operations manager owns it, here is the one-page scope and the list of tools" is an answer that ends the conversation. "We don't really use it" usually turns out to be untrue, and is worse. Will my business insurance ask about AI and what to tell your bank about AI cover those conversations. After a deal, the question gets sharper; the post-merger integration checklist puts it alongside the rest of the first year.
Starting this week
Day one: ask your team who already uses AI tools, for what. Write down the answers.
Day two: pick the person others go to. Ask them, do not assign them.
Day three: write the one-page scope together, with the data sentence and the never-alone list.
Day four: tell the company who owns it and what the rule is, in one short message.
End of the month: a thirty-minute review with the owner: what is in use, what was approved, what went wrong, what comes next.
Picture it going wrong
A company makes its IT contractor "responsible for AI" in a one-line email. Nobody writes down what that means. Six months later, a staff member pastes a customer's contract into a free tool, the customer finds out, and the contractor, the manager and the owner each assume someone else was supposed to have stopped it. A one-page scope and a named person would not have prevented every mistake. It would have made it clear whose job it was to set the rule.
Mistakes that repeat
Making ownership a committee. Giving it to someone with no time. Handing it to an outside vendor who sells the tools. Writing nothing down. Letting the named person make decisions that commit the company.
Frequently asked questions
Who owns AI in a company?
One named person owns the day-to-day, with a one-page written scope. The owner keeps three decisions: what data never goes into outside tools, what a tool never does alone, and what gets built or bought.
Should I hire a chief AI officer?
Rarely first. Most companies of ten to two hundred people need a named person with a few protected hours, not a new executive. Fractional chief AI officer covers when outside help fits.
Should IT own AI?
Sometimes, but AI is mostly a question of how work is done and what data is shared, not of servers. The best owner is usually the person already using the tools well, wherever they sit.
Who decides which AI tools staff may use?
The named person approves tools against the data rule; the owner decides anything that costs real money or gives a tool access to company data.
How long does it take to set up?
A week: find the person, write the one-page scope, tell the company, and set a monthly review.
Is this legal advice?
No. Ask your own adviser about anything touching employment, contracts or data protection.
Not sure where you stand? Take the 3-minute AI readiness scorecard.