Kevin SuraceKevin Surace is an AI pioneer, inventor, author, and human effectiveness authority helping people thrive while AI reshapes work and life.

Why Employees Sabotage AI Rollouts

22 Jun 2026 · 1 min

Watch on YouTube

1:04

In this monologue, Kevin Surace cites a survey indicating that 31% of employees actively sabotage AI rollouts, contributing to an 80% failure rate for AI initiatives in large enterprises. He contrasts this with small companies, where employees are typically hired for an AI-first mindset and thus less likely to resist. Surace explains how AI is transforming software quality assurance: tools like AppVance can generate thousands of test scripts in an hour, uncovering bugs that have never been seen before. This dramatic productivity boost threatens traditional manual testing roles, leading some QA staff to sabotage the AI systems—either by undermining the AI’s output or by deliberately inserting malicious code into AI‑generated scripts to make the system appear faulty. He emphasizes that such human interference is a real, observed phenomenon that hampers AI adoption in big organizations.

Chapters

There was a good survey done that 31% of employees are sabotaging AI rollouts, and that is leading to 80% failures across the board of AI rollouts in large companies. That doesn't happen in a small company because when you join the small company, you're there to be AI first or you wouldn't have joined, right? But in a big company, if you are in software QA and your job was to manually test software that was feature upgrades, bug fixes, et cetera, et cetera, and now AI comes in, writes literally, I mean, AppVance does this, writes literally 1,000 scripts in an hour, test scripts, Runs it and finds bugs you've never found in the history of your company on the production system. You look at your job and you go, who's ever going to pay me to do this very slow work where I test a few things an hour when it just tested 1,000 and it did it for a dollar? So then you sabotage it. We've actually seen this. It's fascinating. Literally have seen people, you know, humans put malicious code inside AI-generated code and say, look, see, it doesn't work. It generates junk. I said that a human added that. I can see what the AI system didn't. No, we didn't. Yes, a human added that.