Hiding Your AI Use at Work Could Cost You the Promotion

Four in ten Gen Z workers are hiding their AI use from their employers, according to a survey reported this week. Half say they feel guilty using it at all.

Now look at the job postings. AI appears in 57% of Bay Area tech listings. Mentions of AI skills in cybersecurity postings have doubled. Demand for AI skills is spreading into roles that have nothing to do with tech. The market is paying for the exact thing a big share of young workers are keeping secret.

We'd treat this as a career problem before a moral one. Every hour of AI-assisted work you hide is an hour that never makes it into your performance record, your promotion packet or your interview stories, and it never reaches the mental file your manager consults when a reorg forces staffing decisions. You keep all the downside of the tool and give up the credit.

Employers ask for AI skills while workers hide them

The demand side is hard to miss. The Jerusalem Post reported surging demand for AI skills in nontech functions like marketing, finance and operations. Inc. argued this week that for many technical roles the strongest candidate may not be the computer science major, and may instead be the person who can apply the tools to a problem in their own field.

The supply side looks very different. Workers describe FOBO, a fear of becoming obsolete, and MarketWatch reports that most of Gen Z now believes AI will take their jobs. The response that seems sensible from inside that fear is to use the tools privately and hand in the output as if you'd done it unassisted. So employers say they want AI-capable staff, and their staff hide the evidence that they are AI-capable.

This may explain part of why the workforce data keeps contradicting itself. The Free Press asks where the AI jobs apocalypse is, Upwork's CEO insists it isn't happening, and the Center for Data Innovation says we need better data before anyone draws conclusions. Some of that measurement problem is behavioral. We can't say how much, but a real share of AI adoption is happening off the books.

The fear has a basis

People aren't hiding their use for no reason. Monday.com became the latest company to blame layoffs partly on AI, joining roughly twenty others tracked this year, and Robinhood's note explaining its 10% cuts drew scrutiny because the AI framing looked like cover. If your employer publicly ties AI to cutting headcount, telling your manager that AI does 40% of your work can feel like drafting your own severance memo.

The second reason has nothing to do with layoffs. Plenty of teams adopted the tools without ever saying what acceptable use looks like, so people are left guessing whether AI-assisted code review, first-draft documentation or synthetic test data is fair play or cheating. Guilt fills the space where a written norm should be.

The bet behind hiding it is that you keep your productivity edge private and stay safe. We think it usually goes the other way. You absorb the higher output expectation, you carry the risk if an unverified error slips through, and you bank none of the credential value hiring managers are now screening for.

Hidden work doesn't get promoted

Internal mobility is where it costs you first. Promotion committees and staffing decisions run on documented scope, and a manager who doesn't know how you got your results will credit your effort instead of your capability. When the next platform team, AI enablement group or automation project gets staffed, the people picked are the ones with a visible track record.

Then there's the output ratchet. Triple your throughput in secret and your baseline resets at the new level, your team plans around it, and you've gained nothing you can negotiate with. The same work, disclosed and written up, is an argument for a bigger scope, a new title, or a move to a team that's actually growing.

Companies are formalizing this, too. Workday has set up a research team focused on AI agent issues, and Cisco's AI Workforce Consortium is building cybersecurity workforce standards. As governance hardens, undocumented shadow use goes from an awkward secret to a compliance problem with your name on it.

How to describe AI-assisted work without overselling it

Disclosure is a craft. The version that helps you describes the outcome, the judgment you applied and the check you own. The version that hurts either hides the tool completely or lists tool names as if knowing a logo were a skill.

A test we like: could a skeptical hiring manager tell from your description what you would have caught that the model got wrong? That's the difference between someone who writes prompts and someone who's accountable for a system. Research software engineers who bridge biology and computer science are a good model here, since their value comes from domain judgment layered on top of the tooling.

• Lead with the result, then the method: "cut incident triage time 30%, using an agent workflow I built and a review gate I own."
• Name the failure you designed around, like hallucinated dependency versions or stale data in retrieval.
• Put a number on the human step: how much of the output you review, and what you reject.
• Skip the tool inventory. When AI shows up in 57% of postings, nobody's hiring for logo recognition.

In interviews, a vague answer costs points

The same gap changes how you should search. Automated screening is under legal pressure (a job seeker is suing Eightfold AI over its automated screening), but that won't slow the filters down this quarter. Referrals are still the reliable way around them, and a referral lands much better when the person vouching for you can describe a specific AI-assisted result.

Expect interviewers to ask how you use AI in your work, and treat a vague answer as points lost. Candidates who can talk about their verification habits, cost tradeoffs and times they overruled a model are beating candidates with better pedigrees. It's the same thing Inc. was getting at about the CS degree.

At the offer stage, TechInformed reported that smaller firms say the layoff wave has made hiring easier, so mid-market companies and scale-ups can now reach people they couldn't before. That's bargaining power if you can show you cover more than one person's worth of work. Bring documented throughput numbers into the pay conversation, because saying you're AI-savvy won't move a band.

If you do one thing this month, get your employer's actual policy in writing. A two-line answer from your manager or your security team turns a private risk into sanctioned practice. If there's no policy, propose one for your team, which is visible work in itself. Then start a log of AI-assisted work with before-and-after numbers for time, defect rate and volume. Mention one AI-related result in each of your next few status updates, and brief one potential referrer on a specific result so they can repeat it accurately.

Work you can't show at review time or in an interview might as well not have happened.