Employees Overstate AI Use at Work to Impress Managers

What happened

Around half of workers say they have misrepresented their use of artificial intelligence at work, either by overstating how much they rely on AI tools or by overstating their expertise with them, according to a finding reported by TechRadar. The most common worry behind the behaviour is the fear of losing their jobs to the technology.

The reporting frames the issue as a gap between how employees present their AI skills to managers and how they actually use these tools day to day. The pattern points to pressure to appear AI-capable, not simply to a lack of AI adoption.

The material available for this briefing is limited to the TechRadar report and its summary. The survey's sample, methodology and fieldwork dates were not included in the supplied material, so readers should check the original research before citing specific details.

Why it matters

For leaders responsible for AI rollouts, this moves the question from tooling to trust. When staff feel they must perform AI competence, managers receive inflated signals about adoption. Decisions on training, licensing and workflow redesign built on those signals may target the wrong problems and miss where capability is genuinely weak.

Job security sits at the centre of the story. Concealment tends to follow fear, and an AI programme that leaves people guessing about their future role will generate exactly this kind of defensive behaviour. Digital transformation that treats employee experience as an afterthought risks undermining the productivity gains it was meant to deliver.

The Renascence take

Most readers will take this as a story about employee dishonesty. We see it as a design problem. When admitting to using AI feels risky, the organisation has built a system that rewards looking capable over learning openly.

Most organisations treat AI concealment as a character issue to be managed with policy reminders. It is better read as a signal about the rules of the environment. Behavioural economics would predict exactly this response when the cost of disclosure is high and the reward for appearing fluent is immediate. Operators who want honest data on AI use should make disclosure low-stakes, reward people for sharing what did not work, and state plainly how roles will change before anyone is asked to demonstrate expertise.