What employees are quietly worried about when AI shows up at work
Not guesswork: what a national survey of employees actually says they worry about, and what a sensible employer does with that.
Worth asking, rather than assuming
It is easy to assume you know what your team thinks about AI at work. Acas asked 1,023 employees across Great Britain directly, in fieldwork carried out between 27 March and 1 April 2025, and published the results on 28 April 2025. The answers are worth reading in full rather than summarised down to a single number.
The actual numbers
Just over a quarter, 26%, worried AI would lead to job losses, the single largest concern. 17% worried about AI simply making errors. 15% felt regulation was not keeping pace. 11% raised data protection concerns specifically. A smaller 5% pointed to environmental effects. Genuinely reassuring, in its way: 17% said they had no concerns at all, and 8% said they did not know.
Employers are more optimistic than staff are anxious
A separate Acas survey of 1,015 senior decision-makers, carried out over roughly the same period, found 35% expecting AI to boost productivity and only 11% seeing no benefit at all. Put the two surveys side by side and a gap opens up: the people deciding to bring AI in are considerably more confident about it than the people who will actually be working alongside it.
Neither figure is wrong exactly, they are simply answering different questions from different vantage points. A decision-maker weighing up a tool sees the efficiency case. Someone whose day-to-day work the tool will change sees uncertainty about what happens to their role. A business that only listens to the first group is missing half the picture of how a rollout will actually land.
What the job-loss figure is probably really measuring
It is tempting to read the 26% figure as pure fear of the technology itself. The more useful reading, and the one this kind of survey usually supports, is that much of that anxiety is really about not being told anything. A tool introduced with no explanation of what it does, why it is there, or what happens to the roles around it will read as a threat almost regardless of what it actually does.
The 8% who said they simply did not know is worth reading the same way. That is not indifference. It usually means nobody has actually explained what is being introduced clearly enough for someone to form a view either way, which is its own kind of gap for an employer to close.
Employees are not primarily afraid of AI. They are afraid of not being told.
What actually reassures people
The things that come up again and again in this kind of research are ordinary: being told plainly what a tool is for before it arrives, being consulted rather than presented with a fait accompli, knowing there is a real person to ask a question of, and the tool itself simply being accurate enough that people are not quietly correcting it behind the scenes.
A worked example
A small retailer introduces an AI-assisted scheduling tool with no explanation beyond a system notification. Within a week, half the team assumes it is there to cut hours. The same tool, introduced with two sentences explaining what it does and does not decide, plus an open session for questions, tends to land as a practical change rather than a threat. The tool did not change. The explanation did.
Why this connects to the worker engagement principle
This is not a coincidence. It is the same reasoning behind the "worker engagement" principle in the government's workplace monitoring consultation and the human-intervention safeguards in the automated decision-making law, both covered earlier in this series. Every one of these separate pieces of guidance and law converges on the same practical instinct: tell people what a system does, and give them a real route to ask about it, before it starts making decisions that touch them.
Where a clear record helps with trust, not just compliance
Jamie HR keeps policies, announcements and acknowledgements in one place, which matters here for a reason beyond paperwork: if staff can see plainly what changed and when they were told, that in itself does a good part of the reassurance work.