The 29% Aren't Sabotaging Your AI. Your Program Has No Clothes.

The stat made the rounds for two weeks. 29% of employees admit to sabotaging their company's AI strategy. 44% among Gen Z. LinkedIn ran with it. Fortune ran with it. Fast Company ran with it. The framing was obvious. A generation of workers, afraid for their jobs, undermining the rollout.

That wasn't the headline.

The same survey said 75% of C-suite executives admit their own AI strategies are "performative theater lacking real internal support." 67% believe their company has already suffered a data leak from unapproved AI tools. 36% lack any formal plan for supervising the AI agents they have already deployed. 35% say they could not pull the plug on a rogue agent if one went sideways. 54% say adopting AI is "tearing their company apart." 73% of CEOs report stress or anxiety related to the thing they are building.

And then 60% of companies plan to lay off employees who won't adopt those strategies.

Read those numbers together. Then tell me who is sabotaging whom.

The emperor has no clothes, and he's blaming the crowd

Most of the 29% isn't sabotage. Most of it is people noticing.

That is not the same thing. Some of the resistance is real. Some workers are genuinely afraid and acting out by feeding company data into public tools, or refusing to touch the sanctioned ones, or quietly routing around the whole program. That exists. I am not going to pretend it doesn't.

But when 75% of the executives rolling the program out already know the program is theater, and a third of them cannot supervise the agents they bought, and two-thirds know data has already leaked through the shadow tools their own teams are using, the sabotage framing collapses.

The workers can see the program is theater. The executives admitted it on a survey. The only person still pretending the suit is there is the one wearing it.

Calling the people who can see the emperor "saboteurs" is a tell.

The sabotage stat is a diagnosis, not a defense

There is nothing about the 29% figure that should make a CIO or CISO feel better about their program. Nothing.

If the number is high in your company, it is not evidence that you are doing the right thing and the workforce is broken. It is evidence that a third of your people are scared or disenfranchised enough to take action against a strategy their leaders are rolling out. That is a signal. A loud one.

The right response is not to cite the stat in your board deck and frame it as an adoption headwind. The right response is to ask why a third of your people do not trust the program.

People matter. They always will. You can swap that for any other framing you want. "People are your most important asset." "Culture eats strategy for breakfast." Pick your flavor. The mechanic is the same. If your workforce is scared enough of your AI rollout to sabotage it, the rollout has a trust problem, not a compliance problem. You don't fix trust with a layoff threat.

And a layoff threat is exactly what 60% of companies have teed up.

About that layoff threat

Here is the part that does not get said out loud often enough.

If your position is "we will replace anyone we can with AI to cut cost," you have placed yourself on that same chopping block.

How long do you think it takes to build an AI that acts like a short-sighted executive willing to make cost-cutting decisions "for the bottom line?" That is a small model. That is a prompt. That is maybe a weekend of fine-tuning. Pattern-matching on quarterly results and issuing layoff memos is not a hard job to automate. It requires opinions and confidence, both of which current models produce cheaply.

Compare that to the AI system that designs the CNC tool path for a jet engine bracket, or runs the actual CNC, or does the root cause analysis on the line when the part comes out of tolerance. Those systems are hard. They require domain data, physics-aware reasoning, tight feedback loops, and tolerance for real-world consequences.

The $500,000 C-level executive is easier to replace than the $90,000 machinist.

Read that sentence again.

If the honest answer is "we will keep people as long as they add value AI cannot replicate," that answer applies both ways. The C-suite does not get to exempt itself from the logic it is using to threaten everyone else. If it does, the workforce is going to notice that too. Some of them already have.

The tech-bro problem

The other half of the story is who is selling you the narrative in the first place.

The loudest voices on AI's impending workforce impact are, with striking consistency, the people selling AI. Foundation model vendors. Cloud hyperscalers. Enterprise AI platforms. Consultancies with an AI transformation practice. The doom forecasts and the product demos come from the same mouths, at the same conferences, in the same week.

That does not make the forecasts wrong. It makes them conflicted.

Before you quote a CEO warning about AI eliminating 40% of entry-level knowledge work, ask two questions.

First. Who has the most to profit if this is true? If the answer is "the person telling me this," that does not disqualify the claim, but it does tell you to discount the certainty. A vendor predicting massive AI-driven displacement while selling you the AI is not a neutral analyst. They are a motivated seller.

Second. If it is true, is my organization, its people, and its data ready for whatever they are selling, or is there work to be done first? This is the question that gets skipped. Most companies have not seen AI success yet. Not in a sustained, measurable, bottom-line way. The survey numbers say that out loud. 79% of executives struggling with adoption. 54% saying it is tearing their company apart. If the average organization is not ready to run the AI they already bought, the answer to "should we buy more" is not automatic.

Buying the next thing is easier than fixing the last thing. It always has been. That is the tech-bro pitch at the heart of every hype cycle. AI just happens to be the current one.

What a CIO or CISO should do Monday morning

Three things. Not five. Not a framework. Not a maturity model. Three.

One. Be honest about your program. What is it for? What is the intended impact on your workforce? If the intent is cost reduction through headcount, say that to yourself before you say it to anyone else. If the intent is capability expansion, commit to it and measure it. If you are in the 75% who know the program is theater, name that internally before you blame the people who can see it. Your staff already knows. The only question is whether you are willing to be honest about it.

Two. Stop quoting the sabotage stat. It does not help you. It does not help your board. It does not help your workforce. If a third of your people are acting against the program, the number is a symptom of a trust failure, not a talking point. Using it as a talking point is how you make the trust failure worse. People matter. They always will. They notice when they are labeled as the problem by the people responsible for the problem.

Three. Ask the two questions on every AI claim. Every vendor forecast. Every executive doom narrative. Every "the industry is moving to X" pitch. Run it through both filters. Who has the most to profit if this is true? And if it is true, is my organization, its people, and its data ready, or is there work to be done first? Most of the noise does not survive those two questions. The signal that does survive is worth paying attention to. That is where your investment should go.

None of this is a rejection of AI

None of this is a rejection of AI.

It is a rejection of the story that has been told about AI for the last two years, in which the executives building the programs are the honest brokers and the workers living under the programs are the problem.

The survey says otherwise. The executives said so themselves.