Last week a UK bank announced it was cutting around 5,000 jobs, and the headlines did the rest. Every outlet I looked at led with AI as the cause (”increases reliance on AI”, “AI-led efficiency drive”), and what most people will have heard was “AI replaces 5,000 jobs”. Which would be horrifying, if it were true.


My first reaction, was “Blimey, that must be some AI they have! Everyone else is struggling to make AI pay for itself, never mind do the work of 5,000 people! - Where can I get me some of their ‘special’ AI? It must be amazing!”


Gartner’s survey of 350 executives, published this spring, found that the firms which had cut deepest weren’t seeing better returns than the ones which had cut least. So either somebody out there has an AI that’s way better than mine, or the story is a bit more complicated than the headline. Which, once I’d read it, it was.


This is what they actually announced - the bank’s finance chief described the cuts as partly about simplifying the organisation and partly about some of the early impacts of AI. The person running the bank called AI “the icing on the cake, not the cake”. A significant share of the roles sit in offshore back-office operations. And a fifth of them had already gone before anyone announced anything.


None of that made the headline, because “partly simplification, partly some early AI effects, a lot of it offshore” doesn’t fit in a headline. “AI replaces 5,000 jobs” does.


I’m not suggesting anyone lied. The bank said AI was part of the picture, and it is. What happened is what always happens: a complicated cause gets squeezed into a simple one liner because a) the simplest version of the story is always better for the audience and b) clicks. It’s the same squeeze that turns careful conversations about AI risk into “there’s a greater than 10% chance that AI will kill us within the next decade.”


I wouldn’t mind but it’s hardly an isolated incident, and it has a name: AI washing, putting an AI label on ordinary business decisions. Amazon’s job cuts last autumn arrived with a staff note framing them around the AI era. Days later its CEO told analysts they weren’t really about AI, not right now, but about culture. Australia’s biggest bank said a voice bot had cut call volumes and let 45 people go; the union said volumes were rising, and the bank reversed the cuts and acknowledged its assessment hadn’t been thorough enough. The union’s verdict was that it had been caught dressing up job cuts as innovation. Klarna told the world AI was doing the work of 700 customer service agents, then admitted it had put cost ahead of quality and started hiring humans again.


And this week brought a fresh one, from Microsoft. It’s offering voluntary exit packages to around 60,000 people, roughly a quarter of its workforce. Note the word offering: that’s how many are eligible, not how many will go. The published version of the memo to staff talks about flexibility, adapting to changing needs and giving people more ways to shape their next chapter. It doesn’t mention AI once. The reporting did, but not in one voice. Some headlines linked the offer to the enormous sums being spent on AI infrastructure, reportedly about $190 billion this year, which makes AI the bill rather than the cause. Others said AI was driving an efficiency push. Same announcement, two different causes.


But I think we need to be careful, because there’s a much bigger issue here than salacious headlines, and it isn’t really about any one company - it’s actually about all of us, and the conversations we’re starting to avoid.


I wrote recently that “the first frame sets the ceiling”. A headline is the first frame for millions of people. Someone reading “AI replaces 5,000 jobs” isn’t weighing simplification, offshoring and a CEO’s cake metaphor. They’re learning that AI is what happens just before the axe falls. Repeat that week after week and you get a public more anxious than the evidence justifies, and an anxious public stops using the technology, stops questioning it, and stops being able to help shape it.


But the damage I’m more worried about is closer to home.


Take the organisations trying to do this properly. IKEA put AI on its customer service line and, rather than lay off the people whose work it took over, retrained around 8,500 of them as remote design advisors. That’s the opportunity cost of automation answered well: here’s the time back, and here’s what we chose to do with it.


In some of my own engagements, we’re already starting to avoid talking to staff about AI, because we’re worried it might make them disengage before we’ve had chance to explain what we’re trying to do to help. Now imagine being the leader in that position. You stand in front of your people to explain how AI could help them, and before you’ve said a word they’re already looking for the axe. They’ve read the headlines. They’ve seen the bank, and Amazon, and they don’t want to be next. When you want to talk about the good AI can do in terms of delivering value for both your customers and your people, it’s becoming harder and harder to even mention the technology, because nobody believes you mean it.


In January, at Davos, Microsoft’s CEO put a name to what’s at stake. Satya Nadella warned that if AI can’t point to something useful in health, education, the public sector and business, it will quickly lose the social permission to keep spending scarce energy on generating tokens. His context was energy and data centres, not headlines or job cuts, but it’s another version of the same issue. Social permission isn’t withdrawn in a single moment. It drains away one anxious person, parent or workplace at a time, and every conversation that never gets started is a piece of transformation that never gets built.


That’s the real cost of AI washing. It isn’t only that some companies are less than straight about why they cut. It’s that the straight ones, the ones doing it the IKEA way, now start every conversation with a trust deficit they didn’t create. And the most valuable thing you have - your people’s ideas about where AI could help - suddenly goes quiet at the very moment you need them to start talking.


So, two asks. If you’re cutting to save money, say you’re cutting to save money. It’s a legitimate decision, it’s efficiency in a challenging economy, and saying so costs far less than the trust you lose when you blame the machine. And if the plan is to use AI to give people time back, say what the time is for, before a headline tells them what their jobs are for.


So consider this a public service announcement. The next time someone tells you thousands of people are losing their jobs to AI, take a look and see if it’s really true. It might be, but I’m betting there’s a real business story underneath the AI icing.