Every CIO is Now a Governance Officer. Most Didn't Apply for the Job.

A few weeks ago, I asked a CIO who runs technology at a mid-size logistics company a simple question: if an AI agent gives one of your customers the wrong answer tomorrow, whose name will show up in the incident report? He didn't pause. "Mine," he said. Not his CTO's. Not Legal. His.

That answer isn't an outlier. It's the majority position of technology leaders worldwide, and it's the starting point for Communications Reckoning: When the AI Agent Fails, Someone Has to Answer - a global research effort we commissioned with CIOs and CTOs across the US, the UK, France, Australia, and Ireland. We wanted to see the impact of AI described by the people running the infrastructure, not the people selling into it.

What we found is a governance gap, and it's opening faster than most boardrooms have noticed.

Accountability has a name, and it's the CIOlink to this section

Let's take a look at the global numbers:

• 52% of those surveyed said CIOs are accountable when an AI agent makes an error in a customer interaction
• 19% say the CTO
• 16% say customer service.
• 6% say legal or compliance.

That 6% number may surprise many because you would think Legal/Compliance exists in large part to own exactly this kind of exposure. This isn't anyone's fault. AI agents moved into production inside the communications stack faster than anyone had built a framework to govern them, and when something went wrong, the person with the technical context to explain it got the job by default. Once AI agents were embedded in the customer experience, they became standard equipment. Accountability didn't get assigned to the CIO, but they absorbed it.

What could be seen to make the accountability to CIOs even more unfair, at least in the eyes of those CIOs, is that not every CIO is necessarily responsible for purchasing communications technology, so they're inheriting someone's decision-making.

Where you sit changes how many CIOs are on the hook. In the US, 58% of CIOs carry this personally, the highest of any market we surveyed. Ireland is close behind at 56%, potentially because it hosts many regulated multinationals, where an AI misstep is also a regulatory event. France sits at 45%, which is well above legal or compliance levels but lower than the global average, because GDPR-era compliance culture has spread technology oversight more formally across functions. The accountability still lands hardest on the CIO everywhere. In France, it just doesn't land alone.

Infrastructure location just became a boardroom questionlink to this section

If you're the one blamed when an AI agent gets it wrong, you need to see what that agent is doing. That starts with knowing where the infrastructure lives.

82% of technology leaders globally say the location of AI infrastructure materially affects which platforms they buy. 30% call it their primary selection criterion. This is a governance requirement now, not a checkbox; the audit trail for a customer interaction lives in the communications platform, not the CRM, not the AI layer bolted on top of it. You don't need to know where the black box sits until the plane goes down. By then, it's too late to ask.

France is the outlier, as only 15% call infrastructure location their primary factor. That's the lowest by a wide margin, but don't read that as indifference because it's actually the opposite and speaks to something specific about the country. In France, data residency is already so embedded in procurement that location stopped being a differentiator years ago. It's already table stakes, which is also why 93% of French technology leaders say data sovereignty has climbed their priority list in the past year, tied with the US for the highest of any market surveyed. It's two different paths, but going to the same destination.

The plan to move is what is stopping migration, not the techlink to this section

Only 9% of CIOs globally say no single platform can meet their organization's requirements. The market solved the capability problem. It hasn't solved the execution problem.

The tractor showed up before the fields were ready for it. Communications consolidation has the same shape. Consider these stats:

• 30% cite migration cost and complexity as the single largest barrier everywhere we looked
• 24% cite regulatory constraints
• 16%, vendor lock-in
• 15% point to internal politics between departments - and that tells you sometimes the harder project is the org chart, not the technology.

Ireland faces the steepest regulatory barrier, at 28%. Migration cost is worst in the US, where enterprise architectures run deep after a decade of best-of-breed buying and bolt-on acquisitions.

CIOs and CTOs are telling us what they need for success, but who's listening?link to this section

Put these three findings together: the person most accountable for AI's mistakes usually has the least unified view of where those mistakes start. Communications infrastructure is the governance layer now, not the plumbing underneath AI, but the place where accountability, data control, and vendor sprawl converge, whether anyone planned it that way or not.

What CIOs need is a single view across everything they've already bought, not more add-ons or tech solutions. Closing that gap is what separates the organizations that get ahead with AI from the ones that get exposed by it.

Most of the industry still sells this in pieces. They see it as a contact center vendor here, a unified communications vendor there, an AI layer bolted on top, and then a systems integrator to make it all behave. That works if you enjoy building a kit car on weekends or stressing out IT teams and partners. It's also a problem if your name is the one in the incident report.

Has anyone asked what happens if this goes all wrong?link to this section

The operational argument for a unified platform has been true for a decade. What's new is that the governance argument is personal now. That's not a warning. It's a description of where the accountability already sits. The only decision left is whether the CIO can see enough to defend it.