AI-Led Layoffs Have a Leadership Problem. And Businesses Are Starting to Pay for It

As companies race to cut costs in the name of AI, many are discovering an uncomfortable truth: automation without thoughtful workforce design creates more disruption than value. The issue isn’t AI itself, it’s how leaders are choosing to implement it.

| Written by Preethy Suresh

The headlines have been relentless. IBM announced roughly 8,000 job cuts in 2024, with HR functions specifically named as targets for AI replacement. Salesforce eliminated 4,000 customer support roles in 2025, with CEO Marc Benioff publicly attributing the cuts to AI-driven productivity gains. Duolingo announced plans to phase out human contractors in favour of automation. Microsoft, Meta, Amazon and Google, the pattern has repeated across the sector, each announcement framed as a confident step into an AI-enabled future. Klarna replaced 700 customer service agents. And the list continues to grow.

But the reversals have already begun. Klarna CEO Sebastian Siemiatkowski later told Bloomberg the company had “focused too much on efficiency and cost. The result was lower quality, and that’s not sustainable.” Klarna has since begun hiring again. though notably through a gig-style model rather than full-time employment. The 700 jobs that were eliminated haven’t exactly returned; they’ve been restructured as contract work, often at lower pay.

Two years ago, that might have been seen as an isolated admission. In 2026, it is increasingly looking like a broader pattern.

The number every boardroom should be paying attention to

Orgvue’s 2025 survey of 1,163 senior business leaders across eight countries found that 55% of organisations that made AI-driven redundancies later admitted they had made the wrong call. Around 34% reported employees leaving as a direct result of how AI was introduced into the business.

Forrester’s Future of Work report predicts that by the end of 2026, nearly half of all AI-related layoffs may be reversed in some form, often through offshore hiring, contract roles or lower-paid positions. And the IBM survey referenced by Fortune paints an even starker picture: only one in four enterprise AI projects delivers the return originally expected.

So why does the pattern continue? Why does every quarter still bring another wave of AI-led layoffs when evidence from previous rounds already suggests caution?

Because, in many cases, the decisions are not really about AI. They are about the optics of AI.

The question isn’t whether AI belongs in the business. It does. The question is whether leaders can distinguish between work that should be automated and work that creates value because a human is doing it.

The mistake hiding behind the AI narrative

Here is what may actually be happening.

A CEO walks into a boardroom just after a competitor announces major cost savings through AI. Pressure builds to deliver a similar headline. A consultant arrives with a presentation deck. A vendor demonstrates a promising new tool. And somewhere in that sequence, a decision gets made about headcount, often before anyone has properly mapped what those employees were actually doing.

The rollout begins before the redesign.

Headcount is reduced before organisations audit where value actually sits within a role. Then, when customer complaints rise, satisfaction scores dip, or service quality declines, businesses are left solving a problem they created themselves, often at a cost that offsets the original savings.

This is not an AI problem.

The technology is, in many cases, doing exactly what it was designed to do.

This is a leadership problem: understanding the difference between deploying a tool and redesigning work around it. And right now, many organisations are still struggling to distinguish between the two.

The contrast leaders should be talking about

US compliance technology company Smarsh introduced an Agentforce-powered service assistant called Archie in 2025. Similar technology category. Similar customer-facing use case. Similar board-level pressure to modernise.

But the outcome looked very different.

Chief Customer Officer Rohit Khanna told CIO magazine that the AI assistant now handles a significant share of customer queries, knowledge-base creation and basic support.

Then came the most important part: “We didn’t let people go.”

Instead, employees were moved into higher-value work, handling more complex issues, escalations and relationship-led interactions where human judgment still matters most. AI absorbed repetitive tasks that were never the best use of human capability in the first place.

Same technology.

Different leadership framework.

Very different result.

The organisations navigating this transition well are not necessarily using more advanced AI. They are doing the foundational work first: mapping responsibilities before making workforce decisions, identifying what can be automated, what should be enhanced through AI, and what depends entirely on human capability, whether that is trust, creativity, judgment or empathy.

The deeper challenge

The dominant narrative today suggests AI is reshaping work.

That may be true eventually.

But right now, what is reshaping work is often the idea of AI, applied too quickly, without structure, and sometimes used to justify cost-cutting decisions that were always about cost.

The technology ends up taking both the blame and the credit for what is, at its core, a familiar restructuring exercise with a new label.

Companies treating AI primarily as a headcount strategy are likely to keep generating reversal headlines.

Those treating it as a redesign opportunity will quietly build stronger operating models.

And the gap between the two is likely to widen over the next 12 to 18 months, because the cost of getting this wrong is no longer theoretical. It is already visible in rehiring patterns, customer experience scores, and institutional knowledge walking out the door, often for good.

The question is not whether AI belongs in business.

It does.

The real question is whether leadership teams can clearly distinguish between work that should be automated and work that creates value precisely because a human is doing it.

If that distinction is unclear, then the AI strategy is not really a strategy.

It is simply a press release waiting to be rewritten.