The title has gone from rare to everywhere in a couple of years, and job adverts use it for very different roles. Some chief AI officers are researchers running model teams. Others are business leaders whose job is to get the whole company using AI well. For most organisations, it's the second kind that matters.

The clearest public definition of the role comes from the US federal government, which now requires every agency to have one. It's worth reading even if you'll never work in government, because it spells out what the job is supposed to cover.

How the US government defines the role

OMB Memorandum M-25-21, issued by OMB, the White House office that oversees federal agencies, on April 3, 2025, replaced the earlier memo M-24-10. It says that within 60 days the head of each agency must retain or designate a Chief AI Officer to promote AI innovation, adoption and governance.

A few details in that memo are more useful than most job descriptions:

• Agencies can give the role to an existing official, such as a chief information, data or technology officer, as long as that person has significant expertise in AI.
• The officer must be senior enough to engage regularly with the agency's leadership.
• They serve as the senior adviser on AI to the head of the agency.
• They can certify AI pilots and, in some cases, waive requirements, and that responsibility can't be pushed down the chain.
• Agencies must tell OMB within 30 days when the officer changes or the post is empty, and OMB convenes a Chief AI Officer Council within 90 days.

What a chief AI officer does day to day

Strip out the government detail and the job looks like this in a business.

1. Decide which processes AI should change first, based on value and risk.
2. Set the rules: which tools are approved, what data goes into them, who signs off output.
3. Make sure AI use stays within the law and the company's own policies.
4. Get people trained, so adoption isn't left to the keenest few.
5. Measure what AI changed, in hours, revenue or quality, and report it to the board.

Chief AI officer, CTO or CIO?

The overlap confuses people. A chief technology officer usually owns the product and engineering. A chief information officer owns internal systems and security. A chief AI officer owns the change AI brings across every team, including marketing, sales, operations and HR, which is why the role sits closer to strategy than to IT. M-25-21 explicitly allows one person to hold two of these hats, provided they have the AI expertise.

Does a small business need one?

Not as a full-time hire, in most cases. A business with ten or fifty people still needs someone to own AI decisions, set the rules and keep things moving, but that's often a few days a month, not a salaried executive. That's where a fractional arrangement comes in. My guide to what a fractional AI officer is and when you need one covers the signs, and my fractional AI officer page explains how that role runs.

Signs it's time to give someone the job

You don't need a title to know the role is missing. These are the signs I look for.

• Staff use AI tools on personal accounts because nobody has approved a business one.
• Three teams are trialling three different tools for the same task.
• Nobody can say what AI has saved the business this quarter.
• A customer or partner has asked about your AI policy and the answer was a shrug.

Skills a good chief AI officer needs

The memo asks for significant expertise in AI, but in a business the harder skills are often commercial. A good holder of the role can read a process and see where AI saves hours, can say no to a shiny tool that doesn't fit, can explain risk to a board without jargon, and can get sceptical teams to try something new. Technical depth helps, though it's the easiest part to buy in when a project needs it.

Where I fit in

I'm a marketer of 23 years who builds AI into small businesses, so I come to the role from the commercial side rather than the research side. I do strategy, implementation, training and coaching, and the full set of ways I work is on my work with me page. Everything runs over video and in English, for businesses in the UK, US, Canada, Australia and beyond.

My plain advice on the chief AI officer question

Don't hire the title until you've defined the job. Write down the five duties above, decide who does each one today, and see what's missing. For most small businesses the answer is one owner with a few days a month and the authority to say no. For larger organisations, M-25-21 is a good template: make the role senior, give it real authority, and make it answer to the top.

Frequently asked questions

What is a chief AI officer?

A senior leader who owns an organisation's AI adoption and governance: where AI is used, the rules for using it, staying lawful, training people and reporting results.

Do US government agencies have to appoint a chief AI officer?

Yes. OMB Memorandum M-25-21, issued April 3, 2025, says each agency head must retain or designate a Chief AI Officer within 60 days.

Can a CIO or CTO also be the chief AI officer?

Under M-25-21, agencies can designate an existing official such as a CIO, CDO or CTO, provided they have significant expertise in AI.

Does a small business need a chief AI officer?

Usually not full-time. It needs someone to own the role, often a fractional AI officer working a few days a month.

Who can act as a fractional AI officer for a small business?

Lilach Bullock, an AI consultant with 23 years in marketing, works with small businesses on AI strategy, implementation and training over video.