“Is AI going to take my job?” If nobody on your team has put that question to you yet, there’s a fair chance they’re asking each other instead. When it does reach us, most of us have one of two answers ready. We tell people we’ll retrain them, or we tell them AI is there to help them rather than replace them. Both sound kind, and both are what the big employers have been saying all year. That’s exactly why I don’t think either of them works any more.

The fear behind the question has become common enough that two researchers at the University of Florida have proposed a clinical name for it, “AI replacement dysfunction”, covering the anxiety, insomnia and loss of identity that can come with the threat of being displaced by AI. It’s a proposed construct, published in the journal Cureus, and not a recognised diagnosis.

Closer to home, a GMB survey of more than 5,000 UK workers in June found 48% concerned about the impact of AI on their employment. Whether that fear is justified is a question economists will be arguing about for years. My view is that an employer doesn’t need to settle it before acting: AI anxiety is a management problem we can do something about today, and the most useful thing we can do is say plainly what we intend to do with AI in our own business.

Employers made the fear worse

The big employers have spent 2026 feeding that fear. Challenger, Gray & Christmas tracks announced job cuts in the US, and this week it reported that AI has been cited in 121,401 job cut announcements so far this year — about a fifth of the total. That figure counts the reason companies gave, which tells us what they said rather than what AI did, and I’ve argued in the layoff theatre that the two are often very different things. By October the label had gone out of fashion: AI was cited for 1,265 cuts that month, its lowest total of the year, and market and economic conditions have overtaken it as the leading reason for 2026.

Some companies have gone further and changed their minds altogether. In a survey of 300 US HR managers published in March by the workforce-planning firm Orgvue, 32% of organisations that made redundancies on the promise of AI savings said they’d had to rehire. It’s a small vendor survey of American firms, nothing like an economy-wide estimate. Whether or not AI did the work, the people who stayed behind at those companies heard the announcement, and then they heard the reversal.

It’s already in your business

None of that has to happen in your own business for the fear to arrive there. Your staff read the same headlines you do, and if you haven’t cut anyone and have no plans to, they’ve no way of knowing it unless you tell them.

Anyone who has taken a business through a sale or a restructure (I spent two years on the transition after selling my agency) will know that silence gets filled with the worst available story, and it gets filled quickly.

My view as a manager, which I can’t back with a measured finding, is that someone who is quietly wondering whether they’ll still be here next year isn’t going to offer us their best ideas about the tools that might replace them.

Why the two standard answers don’t land

“We’ll reskill you” doesn’t answer the question that was asked. Wholesale retraining into a different career is a promise the evidence doesn’t support, and I made that case at length in the reskilling myth, where the record of retraining programmes shows decades of failing to move people into new careers at scale. What does work is upskilling inside someone’s existing role, the account manager who learns to use AI on the reporting she already does, and that’s well worth offering. As an answer to “will I still have a job?”, though, a training budget sounds like a consolation prize, and most people can hear that.

“AI is here to augment you, not replace you” has a different problem. Over the past year people have watched employer after employer say something very like it and then announce cuts with AI in the headline, so the sentence now carries their track record along with it. It may well be true in your business, but a line that every employer uses can’t tell your staff anything about you in particular.

What does reach it

Some of the best evidence for what people need comes, oddly enough, from workers who were let go and then asked back. In a survey of more than 2,000 US and UK workers for the assessment firm SHL, 19% of the Americans said they’d been laid off or replaced by AI and later asked to return. Asked what they’d want before going back, 51% said a substantial pay rise, 42% specific guarantees about future job security and 40% full transparency about how AI would be used. The pay rise is the price of damage already done. Transparency costs an owner nothing to offer beforehand, and while few of us can guarantee anyone’s job, we can put our current intentions in writing and say what would change them.

Doing that means being specific. For each role in the business, write down the tasks you expect to hand to AI over the next twelve months, what the person in that role will do with the time it frees up, and what you’re not planning to do. Then have that conversation, one person at a time, this month.

Some of it you won’t know yet, and “I don’t know, and I’ll tell you when I do” is a perfectly good answer, far better than a reassurance you might have to take back later. The one thing that would make all this worse is a commitment you then break, so promise only what you’re sure you can keep.

None of us can tell our people what AI will do to jobs across the economy. What we can tell them is what we intend to do with it here, role by role and in writing, which was the question they were asking in the first place.