The layoffs came first, and the press releases were straightforward: artificial intelligence would handle the work, and headcount would shrink accordingly. Eighteen months later, a different pattern has emerged in the data.

Robert Half found that 29% of companies that laid off workers after implementing AI have already rehired into those same roles. Forrester’s 2026 Future of Work report puts the number of employers who now regret AI-driven layoffs at 55%, and projects that half of all AI-related job cuts will be reversed in some form by the end of the year. Gartner’s forecast runs further out, estimating that half of companies that cut customer service or operations roles for AI will be restaffing them by 2027, often under new job titles.

Klarna has become the industry’s most-cited example. The company’s AI customer service chatbot was doing the work of 700 employees and was projected to add $40 million in annual profit. What got less coverage was what happened next: the chatbot could handle volume, but the judgment calls that turn a bad customer experience into a retained one kept surfacing gaps. Klarna began rehiring and reached break-even for the first time in May.

The pattern has a name now, the AI boomerang, and for people who’ve spent years inside enterprise hiring, it isn’t much of a surprise.

AI Handles the First Mile. The Last One Still Needs a Person.

“What’s happened is a pretty clean overcorrection,” said Grady Gardner, General Manager and Chief Revenue Officer at talent platform Braintrust. “AI is genuinely excellent at the first mile of most jobs, the high-volume, repeatable parts, and it’s the last twenty to thirty percent, the judgment calls, the escalation, reading a room, where it consistently comes up short. Companies that cut headcount assuming AI covered the whole job are now finding out which mile they eliminated.”

That framing lines up with the pattern showing up across sectors. The roles coming back aren’t the ones AI failed at entirely. They’re the ones where a model can execute the routine 60 to 80% of the work but can’t be trusted with the remainder without a person in the loop, which means the company still needs the role even though it needs less of it done manually.

Gardner argues the more expensive mistake isn’t the layoff itself, it’s what companies optimized for when they made it.

“The initial pitch to the board was almost always cost reduction, replace people, shrink the line item,” Gardner said. “The return this technology delivers shows up as revenue per headcount. A team that’s twenty percent more efficient because AI took the repetitive work off their plate can outperform a team that’s simply smaller and missing the judgment it needs to do the job well.”

The Rehire Is Coming Back More Expensive Than the Layoff

The reversal isn’t cost-neutral. Emerald Book’s reporting on the trend found returning “AI-native” roles, positions that now require managing, auditing, or prompting the AI system that was supposed to replace the function, commanding 20 to 35% higher salaries than the roles they replaced. One-third of employers in a Careerminds survey said they ultimately spent more on restaffing than they had saved through the original layoffs.

That’s the part of the boomerang that doesn’t show up in the initial press release. Recruiting, onboarding, and lost institutional knowledge all carry a cost, and companies that treated the layoff as reversible learned otherwise once the specific people and context were gone.

“Rehiring comes with its own cost structure,” Gardner said. “The person you laid off has often moved on, or the market’s shifted and they cost more to bring back, or the institutional context that made them fast at the job left with them. Companies are relearning that judgment and context aren’t inventory you can restock on demand.”

The Next Version of This Problem Is Already Forming

While the boomerang plays out on the full-time side, a parallel shift is happening in how companies access talent at all. Upwork’s 2026 Future Workforce Index found that 38% of skilled U.S. knowledge workers now freelance, up from 28% a year earlier, and that 58% of full-time employees say they’re actively considering the switch, up from 36%.

Gardner sees the two trends as connected. Companies that over-rotated on full-time AI-driven layoffs are simultaneously watching their best talent shift toward independent, project-based work, which changes what the rehire even looks like.

“The workforce a company rehires into this year looks pretty different from the one it laid off in 2024,” Gardner said. “More of the people they need are choosing not to be full-time employees anywhere, and the company that figures out how to responsibly build a workforce out of both employees and vetted independent talent is going to out-hire the one still trying to solve this with a single headcount line.”

That leaves enterprise hiring teams navigating two problems that used to be separate: figuring out which roles needed a human all along, and figuring out how to bring one back in a labor market that increasingly doesn’t want to be hired the old way.