A buy now pay later fintech's quiet reversal shows the pattern: automate the easy work, dispatch the rest to contractors paid per task. Researchers say the same playbook is now reaching writers, paralegals, and coders.
In 2024, Klarna, the buy-now-pay-later fintech, cut hundreds of customer-service jobs and put an AI chatbot in front of customers. A year later, after complaints about degraded service, the company quietly rehired the humans, this time not as employees but as gig contractors sourced through an outside agency, paid per task, with no full-time status. Klarna's chief executive Sebastian Siemiatkowski called the arrangement "an Uber type of set-up" in a February 2025 podcast, and the phrase is doing a lot of work: it names, in two words, where the rest of the job went The Guardian.
The chatbot handles the easy queries. The contractors handle everything else, and they handle it without the predictability, benefits, or bargaining power that came with the old job. That is the mechanism, and it is no longer confined to one Swedish fintech. The same pattern is now running through delivery, ride-hail, copy work, paralegal research, customer support, and entry-level coding, and the contractors, not the companies, are absorbing the friction.
The scale of the platform layer underneath is harder to see than any one announcement. The International Labour Organization estimates that gig and platform work grew about 90% between 2016 and 2021, and the World Bank now puts the global number of people earning income through labor platforms at up to 435 million, according to figures cited in a May 2026 Human Rights Watch report Human Rights Watch, "Algorithms of Exploitation". The European Council puts platform work in the EU at roughly 43 million people in 2025, a 52% jump since 2022, and country-level counts compiled by HRW from secondary trackers include about 1.2 million platform workers in Mexico, 1.5 million in Kenya, and around 12 million in India.
The human cost is already on the record. HRW's report documents delivery workers in several countries describing up to half of their working time as unpaid waiting, and the same workers say they have no coverage when they are injured on the job. One case the report follows closely: a Beirut Uber driver carjacked while on a trip, left with no workers' compensation, no insurance through the platform, and no sick leave. The platform's algorithm kept dispatching rides to the app on his account during his recovery.
The newest edge of the pattern is in knowledge work, and it is moving faster than the labor statistics. An Upwork survey cited by the Guardian puts the current US freelance and gig workforce at roughly 60 million people, about 39% of the workforce, and Statista projects that share to reach roughly 86 million, or about half, by 2027; both figures are estimates and projections, not measured counts. The largest and fastest-growing segment is no longer drivers and couriers. It is customer-service agents, copywriters, financial analysts, paralegals, writers, and coders, the kind of work a knowledge-sector reader is probably doing right now.
A May 2026 peer-reviewed study by Alexandrea Ravenelle, a University of North Carolina at Chapel Hill sociologist and the author of "Hustle and Gig," tracked creative workers pushed into gig arrangements by AI adoption. The participants include an "algorithmic composer" producing musical loops to train AI systems, a writer evaluating AI-generated text for a major technology company, and an actor doing background work the platform intends to automate. The last of those, quoted in the Guardian, said: "I feel like the construction worker laying the bricks for the gas chambers." Ravenelle's broader point, also quoted in the piece: "I don't believe there's any industry that's safe from this."
The people who study this for a living describe the moment as a design choice, not a law of nature. Mary Gray, a senior principal researcher at Microsoft Research and the author of "Ghost Work," put it this way: "There's no evidence that jobs go away, but there is a lot of evidence that as soon as you can dismantle full-time employment, companies will do that." Lena Simet, a Human Rights Watch senior adviser on economic justice and a co-author of "Algorithms of Exploitation," framed the same point from the other side: "What we're seeing in gig work is in some ways the first indication of something broader... So many jobs could be gigified."
Ravenelle's sociological shorthand is that work has been moving "from the career to the job to the gig," and AI makes the next step easier. The reason it is easier is not a mystery: a model can be trained to handle a defined task, the task can be priced, the task can be dispatched to a contractor, and the contractor can be replaced when the model improves. None of that requires a layoff announcement. It looks, from the outside, like ordinary product work.
That is the constructive part of the story, if constructive is the right word: none of this is fully baked. Klarna made a public reversal, and the contractors it hired are still inside that arrangement. The HRW report is a May 2026 publication, which means the case studies and country counts are the strongest evidence on the table but they are also recent, and the regulatory response in the EU, the United States, and the larger platform economies is still in motion. The next concrete thing to watch is whether the contractors being hired in 2025 and 2026 stay contractors, or get rolled back into payroll the way Klarna's last batch did. The answer to that question is being set, right now, by the platforms that broker the work and by the companies that buy it.