1. Blog
  2. Productivity
  3. They Replaced Staff With AI. Now 3 in 10 Are Bringing the Jobs Back.

They Replaced Staff With AI. Now 3 in 10 Are Bringing the Jobs Back.

Sources checked 10 Nov 2026 Productivity
They Replaced Staff With AI: a worker walks back through a lit doorway as an AI figure steps away from the desk
Table of contents
  1. Are companies really rehiring after AI layoffs?
  2. Why are companies bringing the jobs back?
  3. What does it cost to get this wrong?
  4. Is AI still taking jobs?
  5. What should a small business learn from AI rehiring?
  6. When is replacing a role with AI the right call?
  7. Where does a virtual assistant fit in?
  8. Frequently asked questions
  9. The bottom line

Some of the companies that swapped people for AI are now writing the same job ad again.

Quick answer: Yes, companies are rehiring after AI layoffs, but it’s a partial reversal, not a stampede. Robert Half found that 32% of US employers that eliminated positions after adopting AI later added those roles back, most often because the work needed more context, relationships or checking than expected. Employers still cited AI in 121,401 announced US job cuts through October 2026. For a small business, the lesson is to automate tasks first and change a role only after AI has proved it can do the work.

A quick disclosure: I run a virtual assistant company, so I have a stake in people still being useful. I’ve also worked as a VA for 16 years and watched AI take real work off the table. So I’ll show you the numbers that cut against my side too.

Are companies really rehiring after AI layoffs?

Yes, some are. But the surveys behind the headlines measure different things, so it’s worth seeing them side by side.

SourceWho was askedWhat they foundWhat it really tells you
Robert HalfUS hiring managers whose organizations cut positions after adopting AI32% later added those roles, or very similar ones, backHow often employers reverse the decision
SHL2,036 employed adults in the US and UK, June and July 202619% of US workers and 13% of UK workers report firsthand experience of being replaced by AI and asked backWhat workers say happened to them
Gartner275 customer service leaders, July and August 20267% have rehired after earlier cuts, and 11% plan to within 18 monthsCustomer service only, and still early
ForresterPredictions 2026 report, published October 202555% of employers regret laying off workers because of AIEmployer sentiment, not actual rehiring
Four figures on AI rehiring: 32% of US employers that cut roles after adopting AI added them back (Robert Half), 19% of US workers report being replaced by AI and asked back (SHL), 7% of customer service leaders have rehired after AI cuts (Gartner), and 55% of employers regret AI layoffs (Forrester).
Four surveys, four different questions. Sources: Robert Half 2026, SHL 2026, Gartner 2026, Forrester via The Register 2025.

You may have seen the Fortune headline saying nearly 20% of US workers replaced by AI are being asked back. Read that one carefully.

SHL’s release says nearly one in five US workers report firsthand experience. It doesn’t spell out the base, but it reads as a share of everyone surveyed, not a share of people who lost a job to AI. It’s also self-reported, so it depends on each person’s view of why their job went.

Two numbers compared. SHL: 19% of US workers report firsthand experience of being replaced by AI and asked back, which reads as a share of everyone surveyed. Robert Half: 32% of US employers that cut roles after adopting AI later added them back.
A percentage of whom? That question changes what these two numbers mean.

That’s why I lean on Robert Half’s number. It asks the cleanest question: of the employers who actually cut roles, how many put them back? Just over 3 in 10 did. Robert Half has even named them “AI correction hires.”

Ten people icons with three highlighted, showing that 32% of US employers that cut positions after adopting AI later added them back.
Robert Half calls these roles AI correction hires.

Why are companies bringing the jobs back?

Because the job was bigger than the task list. Robert Half asked hiring managers why the roles returned, and these were the top reasons:

Bar chart of reasons US employers reinstated roles cut after adopting AI: required institutional knowledge or context 40%, involved relationship management 39%, needed as business demand increased 39%, required more human oversight and quality control than expected 38%.
Why roles cut after adopting AI were reinstated. Source: Robert Half, May 2026 Labor Market Update.
  • 40% needed institutional knowledge AI couldn’t replace. That’s knowing which supplier always bills late, or that your biggest client hates phone calls. Nobody wrote it down because nobody had to.
  • 39% involved relationships AI couldn’t replicate. People forgive a mistake from someone they know. They don’t extend that to a chatbot.
  • 39% were needed as demand grew. This one is good news, and it isn’t a failure of AI at all.
  • 38% needed more oversight and quality control than expected. This is the hidden cost. Somebody has to check the output, and that somebody is usually on the payroll or is you.

Klarna is the best-known example. In February 2024, the company said its AI assistant could do the work of 700 customer service agents and had handled 75% of customer chats in its first month. By May 2025, CEO Sebastian Siemiatkowski was telling Bloomberg the company was hiring people again. AI was cheaper, he said, but delivered “lower quality.”

Klarna didn’t abandon AI. It decided customers should always be able to reach a person. That’s a much more useful conclusion than “AI failed.”

Timeline of Klarna's AI customer service. February 2024: AI assistant said to do the work of 700 agents and handled 75% of chats in its first month. May 2025: the CEO says the company is hiring people again. Conclusion: customers should always be able to reach a person.
Klarna’s reversal, from its own statements and the CEO’s interview with Bloomberg.

What does it cost to get this wrong?

More than the salary you thought you saved. When SHL asked workers about returning after being replaced by AI, here’s what came back:

  • 51% would want a substantial pay increase
  • 42% would want job security guarantees
  • 40% would want full transparency about how AI is used in their role
  • 8% would refuse to go back under any circumstances
Bar chart of what workers would want before returning after being replaced by AI: 51% a substantial pay increase, 42% job security guarantees, 40% full transparency on AI, and 8% would refuse to go back.
What it would take to get people back, according to SHL’s 2026 survey.

SHL’s Lucy Beaumont called it “a wake-up call.” In her description, the AI falls short, the gap becomes obvious, and the business rehires the same experts at a higher cost.

Small businesses have even less room for this. In a Robert Half survey of more than 250 US small business leaders, 47% said finding skilled talent is harder than a year earlier, and 56% reported significant skills gaps. If your one experienced admin leaves, you may not get that person, or that knowledge, back.

Two figures from US small business leaders: 47% say finding skilled talent is harder than a year earlier, and 56% report significant skills gaps on their teams.
Why a small business can’t count on simply rehiring. Source: Robert Half.

There’s also an uncomfortable twist, and it touches my own industry. Forrester predicts half of AI-attributed layoffs will be reversed, with much of the work going to “lower-wage human workers, offshore or at lower salary.”

Bar split in half showing Forrester's prediction that half of AI-attributed layoffs will be reversed, with much of the work going to lower-wage human workers, offshore or at lower salary.
This is Forrester’s forecast from October 2025, not a measured result.

Offshore support can be a good decision. I obviously think so. But it should be a decision you make on purpose, with the work properly documented and managed, and not the cleanup after an automation project went wrong.

Is AI still taking jobs?

Yes, and anyone who tells you this is all over isn’t reading the data.

Employers cited AI in 121,401 announced US job cuts through October 2026, according to Challenger, Gray & Christmas. That’s more than double the 54,836 for all of 2025, and about 20% of all announced cuts this year.

Bar chart of US job cuts where employers cited AI: 54,836 in all of 2025 and 121,401 in January through October 2026.
Announced US job cuts where employers cited AI. These are reasons employers state, not an independent measure. Source: Challenger, Gray & Christmas, via HCAMag.

Three things to keep in mind before you panic or celebrate:

  • It’s what employers say. Challenger counts the reason given in the announcement, and “AI” can sound better to investors than “we overhired.”
  • It’s uneven. May 2026 set a record at 38,579 AI-cited cuts. October had just 1,265, the lowest month of the year.
  • Most customer service teams haven’t cut. Gartner found only 21% of service leaders had reduced human agent headcount because of AI. Gartner analyst Emily Potosky said “widespread reductions are not yet evident in the data.”
Bar chart of announced US job cuts citing AI in 2026: a record 38,579 in May and 1,265 in October, the lowest month of the year. Through October, AI was cited in about 20% of all announced cuts.
AI-cited cuts swing widely from month to month. Source: Challenger, Gray & Christmas.

So both things are true. AI is cutting some jobs, and some of those cuts are being undone.

Grid of 100 squares with 21 highlighted: 21% of 275 service leaders reduced human agent headcount because of AI, 7% have rehired after earlier reductions, and 11% plan to rehire within 18 months.
Gartner’s survey of service and support leaders, July and August 2026.

What should a small business learn from AI rehiring?

Replace tasks, not jobs. And don’t change a role until the AI has earned it.

Gartner’s Kathy Ross put it well: “Leaders should treat workforce reductions as an outcome of proven and sustained AI performance.” In other words, the savings come after the proof. Many of the companies now rehiring did it in the opposite order.

Two four-step sequences. Backwards: cut the role, switch on the AI, find the gaps, rehire at a higher cost. In order: run AI beside the person, compare the results, decide who checks it, then change the role.
Prove it first, then change the role.

Before you take a person out of a role, or skip a hire because “AI can do that,” ask these five questions:

  1. What does this person know that isn’t written down? If you can’t list it, AI can’t learn it.
  2. Who would notice if they disappeared? Clients, suppliers and teammates are the real test.
  3. Who will check the AI’s work, and for how many hours a week? Put a number on it. If the answer is “me,” count your own time.
  4. What happens when the AI is wrong? A clumsy internal summary is fine. A wrong invoice sent to your biggest client isn’t.
  5. Has the AI done this work reliably for a few months? Run it alongside the person first and compare the results.
Checklist of five questions to ask before cutting a role for AI, covering unwritten knowledge, who would notice, who checks the AI's work, what happens when it's wrong, and whether it has worked reliably for a few months.
A quick test to run before you remove a person from a role or skip a hire.

Then sort the work by what a mistake would cost you:

Type of workExampleBest setupWhy
Repetitive, low riskMeeting notes, inbox sorting, common FAQsAI on its own, with spot checksErrors are cheap and easy to catch
High volume, errors cost moneyInvoice entry, CRM updates, data cleanupAI does it, a person checks itThe speed is real, and so is the risk
Exceptions and follow-upChasing late payers, rescheduling, missing paperworkA person using AISomeone has to own the outcome
RelationshipsUpset customers, key accounts, renewalsA personTrust is the job
Regulated sign-offTax filings, legal adviceA licensed professionalAccountability can’t be delegated to software
Ladder of five work types by rising cost of a mistake: repetitive low-risk work suits AI alone, high-volume work suits AI with a person checking, exceptions suit a person using AI, relationships need a person, and regulated sign-off needs a licensed professional.
The more a mistake would cost you, the more a person needs to own the work.

When is replacing a role with AI the right call?

Sometimes it is, and you shouldn’t feel bad about it.

It can make sense if the work is genuinely repetitive, the AI has matched human quality over a real test period, someone is clearly responsible for monitoring it, and customers can still reach a person. It also makes sense to change the shape of a role rather than remove it. Gartner found 71% of service leaders are moving agents into new roles within service and support.

Four conditions for replacing a role with AI: the work is genuinely repetitive, AI has matched human quality over a real test period, someone is responsible for monitoring it, and customers can still reach a person.
If all four are true, you have a real case for changing the role.

And if a simple tool covers your needs, just use the tool. I went through exactly when that’s enough in will AI replace virtual assistants.

Where does a virtual assistant fit in?

Every AI setup needs an owner. Look again at that 38%: employers needed more oversight than they expected. In a small business with no IT team, that oversight lands on the founder by default.

Two columns. The AI handles volume, meeting notes and inbox sorting. The person handles checking the output, exceptions and follow-through. 38% of hiring managers who brought roles back said they needed more oversight than expected.
AI brings the speed. Someone still has to own the result.

A virtual assistant who works with AI is a practical way to fill that gap. The AI handles volume, and the person handles checking, exceptions and follow-through. You can apply it to customer service, email management, calendar management or bookkeeping.

It also addresses the institutional knowledge problem, if you set it up properly. At 7VAS, every plan includes a dedicated VA, a trained backup VA, a QC reviewer and an account manager, so what your VA learns about your business doesn’t sit with one person.

Every 7VAS plan includes a dedicated VA, a trained backup VA, a QC reviewer and an account manager. Part Time is US$699 a month for 80 hours and Full Time is US$999 a month for 160 hours.
Four roles behind every 7VAS plan, month to month.

For cost, 7VAS starts at US$699 a month for 80 hours and US$999 a month for 160 hours on our price plans, month to month. That’s more than an AI subscription and well below a full-time hire. Our virtual assistant cost guide breaks down the wider market, and our statistics page has more data on AI use in small businesses.

Frequently asked questions

Are companies rehiring after AI layoffs?

Yes, some are. Robert Half found that 32% of US employers that eliminated positions after adopting AI later added those roles, or very similar ones, back. In customer service, Gartner found 7% of leaders had rehired after earlier reductions and 11% planned to within 18 months, so the trend is real but still limited.

What is AI rebounding?

AI rebounding is SHL’s term for workers being replaced by AI and later asked to return by the same employer. SHL’s 2026 survey of 2,036 employed adults found 19% of US workers and 13% of UK workers report firsthand experience of it.

What percentage of workers replaced by AI get asked back?

No survey has measured this cleanly. The widely shared figure of nearly 20% comes from SHL, whose release says 19% of US workers report firsthand experience, which reads as a share of all workers surveyed. The clearer employer figure is Robert Half’s: 32% of employers that cut roles after adopting AI added them back.

What are AI correction hires?

AI correction hires is Robert Half’s term for positions an employer eliminated after adopting AI and later had to add back. The most common reasons were institutional knowledge AI couldn’t replace (40%), relationship management (39%), growing business demand (39%) and more oversight than expected (38%).

Why do AI layoffs backfire?

AI layoffs backfire when a company automates a job before understanding everything the job involves. Robert Half’s data points to missing context, lost relationships and more checking than planned. Klarna’s CEO said its AI customer service was cheaper but “lower quality,” and the company began hiring people again in 2025.

How many jobs has AI replaced in 2026?

Nobody has an exact count. Challenger, Gray & Christmas reports that employers cited AI in 121,401 announced US job cuts through October 2026, up from 54,836 in all of 2025. That figure reflects the reasons employers give, so it isn’t an independent measure of jobs AI actually replaced.

Do workers replaced by AI demand more money to come back?

Many say they would. In SHL’s 2026 survey, 51% of workers willing to return would want a substantial pay increase, 42% would want job security guarantees and 40% would want transparency about how AI is used in their role. Another 8% said they would refuse to return under any circumstances.

Should a small business replace staff with AI?

Replace tasks first. AI works well on repetitive, low-risk work, but roles that depend on context, relationships or follow-through are the ones employers most often had to bring back. Test AI alongside the person for a few months and decide who checks its work before you change the role.

The bottom line

AI layoffs aren’t over, and AI isn’t going away. But a meaningful share of the employers who moved fastest have already paid to reverse the decision.

You don’t have to repeat their experiment. Automate the tasks, keep a human in charge of the outcome, and let the results tell you when a role should change.

Not sure which parts of your workload belong with AI and which need a person? Have a look at our virtual assistant services and book a free consultation. Tell us what’s on your plate, and we’ll help you sort it.

John Peter D, founder of 7 Virtual Assistant Services
About the author:

John Peter D

Founder & CEO, 7 Virtual Assistant Services

John has worked as a virtual assistant for 16 years and founded 7VAS to give small businesses a dedicated VA with the cover and quality checks he wished his own clients had. His team works from Chennai, India, with Philippines-based VAs and bilingual English–Spanish teams in Brazil and Chile, supporting 200+ clients across the US, UK, UAE, Singapore and Australia.