Many CIOs are now worried about their jobs.
That is what a CIO advisor told me yesterday. The CEO has taken over AI, and the CIOs ability to say no is disappearing. Most of his clients are hearing some version of "find a way to say yes, or I will find someone who can". Unsurprisingly, that's leaving a decent number of CIO's spooked.
Is that one advisors circle, or something wider? I went looking.
BCGs AI Radar 2026 found that 72% of CEOs now say they are the main decision makers on AI, and half of them believe their own jobs depend on getting it right. That explains the pressure.. if your job rides on AI, you are not going to be patient with the person raising risks.
Dataikus CIO survey shows the other side. 74% of CIOs think their jobs are at risk if their company doesnt deliver measurable AI gains, and 62% say their CEO has questioned or challenged their AI decisions.
Cisco's CEO research lists deploying AI agents as a top three priority for 2026, and security and control of those agents as the number one concern. So the CEO worries about the same risk.. they just expect someone else to sort it, quickly. And "sort it" doesn't mean saying NO.
Guess who that someone is? The CIO, who still answers for the data leak and the agent that did something nobody approved. The accountability stays with you.. the authority to say no has gone. A CSO Online piece reports nearly two thirds of senior decision makers already use unauthorized AI tools, so the people overriding policy are often the ones above you.
Saying no was part of the job for good reason, and most of you earned that instinct cleaning up after the last wave of shadow IT. Now imagine telling the CEO the data classification isnt ready when the board wants agents this quarter.
That leaves one safe position.. a yes you can defend. You need to know what AI is running (you cant govern an agent you dont know exists), control what agents can reach, and keep a record of what they did. When something breaks, the first question is who signed off, and the CIOs name is usually on the paperwork.
Lock it down so hard that AI becomes slow and useless, though, and people route around you, or the CEO decides the CIO was the problem after all. A governance layer that makes AI functional but useless fails the same test as having no governance at all.
So the job is a yes that holds up when something goes wrong, and still delivers what the CEO is betting their career on.
CIOs in my network... how much room do you really have left to say no? Id genuinely like to know...
This problem is exactly why Portainer-Command now exists... secure use of Agentic AI Operations harnesses against Kubernetes environments. Just in time, and just enough permissions, all writes redirected to Git, full audit log, instant rollback. (portainer.ai).
The CEO is now driving the AI train
