A year ago, most CEOs still treated AI as somebody else’s project. Hand it to the CTO, check in at the quarterly board meeting, move on. That’s over. According to BCG’s AI Radar 2026, a survey of more than 2,300 executives across 16 markets, 72% of CEOs now say they’re the primary decision-maker on AI at their company – roughly double the share who said the same thing a year earlier.
That’s not a small shift. It means AI stopped being a technology question somewhere in the last twelve months and turned into a strategy question, sitting on the same desk as pricing, hiring, and where to raise the next round.
If you run a company – even a five-person startup – this affects you whether or not you’ve noticed yet. Here’s what actually changed, why it happened so fast, and what it means to be genuinely ready instead of just saying you are.
What “owning the AI decision” actually means
Owning the AI decision isn’t the same as being the person who approved a ChatGPT subscription. It means the CEO is now the one deciding:
- Which processes get automated and which stay human
- How much of the budget goes to AI tools versus everything else competing for that money
- Who’s accountable when an AI system makes a bad call that affects customers or revenue
- How fast (or slow) the company moves relative to competitors already deploying agents
Boston Consulting Group’s framing is blunt about why: AI decisions now touch revenue, customer experience, competitive position, and workforce structure at the same time. No single functional lead – not the CTO, not the head of ops, not marketing – has authority across all four. So it lands on the CEO by default, whether they asked for it or not.
There’s a catch worth sitting with. A separate 2026 survey from Dataiku, based on Harris Poll data from 900 CEOs, found that 70% claim ownership of AI strategy, but only 6% are actually involved in the day-to-day decisions that shape it. Ownership on paper and ownership in practice are not the same thing, and that gap is exactly where a lot of companies are quietly getting exposed.
Why this happened so fast
Three things collided in the last year:
1. The technology crossed from “helpful” to “consequential”: A chatbot that drafts emails is a productivity tool. An AI agent that approves credit, prices a service, or handles a customer complaint end-to-end is making decisions with legal and financial weight. That’s not a CTO-level call anymore.
2. Boards started asking sharper questions: The same BCG research found 61% of CEOs say their boards are pushing the AI transformation faster than the organization is actually ready to absorb. Boards aren’t asking “are we experimenting with AI” anymore – they’re asking what it’s returning.
3. Spend went up, and spend concentrates accountability: Companies in the BCG survey expect to roughly double AI investment in 2026, from about 0.8% to 1.7% of revenue. Once a budget line gets that size, somebody with P&L responsibility has to own it. That’s the CEO, not the IT department.
For founders reading this, the honest version is: you already made this decision the moment you started using AI tools to build product, write copy, or run support. You just might not have called it a strategy yet. Generative AI for business is no longer optional context – it’s the terrain you’re already standing on.
The tension nobody’s talking about enough
Here’s the part that gets skipped in most coverage of this stat: CEOs taking ownership doesn’t mean CEOs feel confident.
The Dataiku/Harris Poll research found 80% of CEOs admit they actively question or challenge the outputs their own AI systems produce, and just over half still require a human to sign off before an AI-influenced decision goes live. Confidence in deploying AI agents at real scale actually dropped, from 41% to 31% year over year, even as most of those same companies plan to push agents into full production anyway.
So the picture isn’t “CEOs are confidently steering the ship.” It’s closer to: CEOs are holding the wheel because someone has to, while privately unsure the ship is built for the water it’s in. If that sounds familiar, you’re not behind – you’re normal. The gap between claiming ownership and actually being ready is the norm right now, not the exception.
Signs your company isn’t actually ready, even if you “own” the decision
A few honest checks, based on where the research says most companies are actually stuck:
- You can’t name who’s accountable when an AI tool gets something wrong: If the answer is “well, nobody really owns that,” you have a governance gap, not an AI gap.
- Your AI spend grew, but nobody’s measuring return on it: Doubling investment without a way to check what it’s buying is how budgets quietly become sunk cost.
- You’re using AI agents in customer-facing workflows you can’t fully observe: If you can’t see what the agent said or did after the fact, you can’t defend it later either.
- Your team adopted tools faster than you built any policy around them: Common in small companies – everyone’s using AI individually, nobody’s agreed on where the lines are.
- “AI strategy” lives in a slide deck, not in how decisions actually get made day to day
None of these are reasons to panic. They’re reasons to close the gap before a customer, a regulator, or a bad quarter closes it for you.
What being ready actually looks like
You don’t need an enterprise AI governance team to close this gap. For a founder or small-business CEO, readiness looks more like this:
Pick your two or three highest-stakes AI use cases and name an owner for each: Not “the team” – one person who can answer for what the AI did if it goes sideways. This alone fixes most of the accountability gap the surveys are picking up on.
Set a review cadence, even a rough one: Once a month, look at what AI touched – pricing, support replies, hiring screens, whatever applies – and ask if it’s actually working, not just running.
Decide where a human stays in the loop on purpose: Not everywhere. Just the handful of decisions where getting it wrong costs you a customer, a hire, or a legal headache.
Write down the one or two rules your team actually needs: Something as simple as “don’t paste client data into a public AI tool” and “a person reviews anything that goes out under the company’s name” covers most early-stage risk.
Revisit the budget conversation with numbers, not vibes: If AI spend is going up, know what it’s replacing or unlocking. That’s the difference between an investment and a subscription nobody canceled.
If you’re still picking the right tools to build this around, our current list of AI tools for startups is a reasonable place to start narrowing the field.
The real question behind the stat
The 72% figure isn’t really about AI. It’s about a leadership job description quietly getting rewritten. CEOs didn’t ask to become the default AI accountability holder – it happened because the decisions got too cross-functional for anyone else to make them alone.
Owning the decision on paper is easy. Boards will keep pushing, and 61% of CEOs already say that pressure has outpaced what their organizations can actually absorb. The founders and CEOs who come out ahead won’t be the ones who adopted AI first. They’ll be the ones who can say, clearly and specifically, who’s accountable for it, what it’s costing, and what it’s actually returning.
That’s the readiness question underneath the headline. Worth asking it before your board – or your customers – ask it for you.
FAQ
What does it mean that CEOs “own” the AI decision?
It means the CEO, not the CTO or CIO, is now the primary person deciding AI investment, adoption, and accountability at most companies – because AI decisions now affect revenue, customers, and workforce structure all at once, which no single functional leader can decide alone.
Where does the 72% statistic come from?
It comes from Boston Consulting Group’s AI Radar 2026 report, published in January 2026, based on a survey of more than 2,300 business executives, including 640 CEOs, across 16 markets.
Is CEO ownership of AI the same as CEO confidence in AI?
No. Separate research from Dataiku and Harris Poll found most CEOs who claim ownership are only involved in a small share of the actual day-to-day AI decisions, and a majority actively question the outputs their AI systems produce.
How can a small business or startup apply this without a big governance team?
Start small: name one accountable owner per major AI use case, set a monthly review of what AI is actually touching, decide where a human stays in the loop on purpose, and write down one or two non-negotiable rules for how the team uses AI tools.
What do you feel about this post?
Like
Love
Happy
Haha
Sad
