Lisa, a freelance graphic designer in Spain, noticed a change in her work after ChatGPT came out in 2022. Instead of the slow trickle of one-off jobs she used to get for creating logos and packaging, she started receiving a flood of requests to fix AI-generated designs—sharpening blurry images for print or turning flawed files into something usable.
By 2025, Lisa, who asked not to be fully named to avoid unwanted inquiries, said that 90% of her incoming logo and packaging requests involved cleaning up AI-generated content. That work made up 60% to 70% of her yearly income. But it was exhausting. Clients lowballed her, some AI designs were so broken she had to start over from scratch, and she worried about being part of AI-related copyright issues. By the end of the year, she started turning those jobs down.
“It just felt so soulless,” said Lisa, who now focuses on “human-made designs.”
Lisa is part of a growing group of creative freelancers hired to clean up “AI slop”—a term for low-quality, AI-generated work. As businesses use tools like OpenAI’s ChatGPT and Anthropic’s Claude to produce marketing copy, illustrations, books, and videos, they’re hiring creative professionals to make AI output sound more human, fix botched images, and salvage flawed footage.
This work acts as a stopgap in a tough job market, helping freelancers who work full-time or part-time make extra money. But clients often don’t realize how much time and skill it takes, expecting quick, cheap fixes when the cleanup can be as demanding as creating something from scratch, freelancers said. While the high volume of cleanup gigs can pay well, freelancers are split on whether it’s worth it. Some find it dull and creatively empty, while others enjoy the challenge. As AI improves, freelancers wonder how long the demand will last and what they’ll do once the jobs disappear.
For now, demand for AI cleanup work is on the rise.
On Freelancer.com, a platform with over 88 million users, job posts tagged with phrases like “correct AI,” “AI hallucination,” and “AI error” jumped 87% to 10,760 worldwide between August 2025 and June 2026, according to internal data shared with the Guardian. Most of these jobs were in graphic design, followed by video editing, proofreading, and content writing. Upwork saw AI remediation gigs rise 70% year over year, and Fiverr reported that searches for “AI cleanup” services grew more than 20 times between 2023 and 2026, based on data shared with the Guardian.
The rise of AI could spell trouble for the creative freelance job market. Listings for freelance jobs at risk of automation fell 21% in the eight months after ChatGPT launched, and image-creation gigs dropped 17% after AI image generators appeared, according to a 2025 study in the journal Management Science. Those losses have continued, per a 2026 study.
At the same time, more workers are turning to freelance work as companies cut staff and hire less, making competition tougher. The average project on Freelancer.com got 54 bids in 2025, up 8% from the year before, according to the site’s annual report.
As companies trim costs, they’re also changing what freelance assignments look like—shifting from creative work to polishing AI output. Matt Barrie, CEO of Freelancer.com, said cleanup work is often commissioned by small companies and entrepreneurs who use AI to create a first draft but then hit problems they can’t solve themselves.
“[AI output] needs a human to turn it into something polished that can actually be used,” he said. The time and money saved on the first draft can be eaten up by the “incredibly time-consuming” work of getting it ready for commercial use.
The rise of the AI cleanup crews
For some freelancers, cleanup is taking up a bigger share of their workload, while traditional creative work is shrinking.Projects are getting harder to find. Todd Van Linda, a full-time freelance illustrator in Florida, started taking on AI cleanup jobs in 2023 as his usual work—like drawing illustrations, video game art, and architectural plans—became less steady. The work usually means removing visual noise, improving resolution, and spotting mistakes like an extra toe. “Money,” Van Linda said about why he does it. “The jobs were there to be done.”
Even though some clients expect AI cleanup to cost less than original work, Van Linda won’t lower his rates. When one potential client offered about $500 to fix 13 to 15 AI-generated illustrations for a children’s book, expecting each to take around 15 minutes, he turned it down. He said projects can take anywhere from a few hours to days, depending on the size. His rate is $65 an hour and “doesn’t go down from there,” he said.
For Kym Dunbar, a freelance writer and editor based in Australia, AI cleanup now makes up about 60% of her workload, as writing jobs like blogs and website copy have dried up. As a result, Dunbar said her monthly earnings from writing original content have dropped to about a third of what she made during the pandemic.
Dunbar got into cleanup work by chance in 2024, when a client asked her to revise an AI-generated article. She quickly learned to spot common issues, from repetitive language and awkward paragraph flow to a lack of emotional depth. Since then, she’s worked on about 40 AI-generated nonfiction books, long articles, and technical papers.
“There’s a lot fewer good projects available for writers and editors these days,” Dunbar said.
But “cleanup” can downplay how much work is actually involved.
Nathan McConnell, a multimedia editor who freelances alongside his full-time job, started editing AI-generated videos in 2023. His work has ranged from removing reflections in AI real estate videos to fixing robotic voiceovers and misshapen 3D models. The requests have grown from two or three projects in 2023 to eight to ten so far this year, he said.
One client brought him about 100 AI-generated images of humanoid animals for a tarot deck, full of extra fingers, two left feet, and other errors. McConnell spent two to three hours fixing each image in Photoshop, expecting the project to take nearly two weeks. But after finishing the first 10 cards, he stopped due to health issues and constant changes from his “persnickety” client.
Not everything that comes to McConnell can be saved, he said. That might include videos where the physics feel wrong, like shadows falling in the wrong direction, or animations that reuse another creator’s work without permission.
“I’ve told a client before that it’s ‘too’ slop,” McConnell said. “There’s no way I can fix this.”
While McConnell and Van Linda would rather create original work, not everyone finds fixing AI outputs discouraging. Dunbar sees it as “interesting” and “challenging,” and feels satisfied turning a weak draft into polished writing.
What bothers her is that people increasingly view AI not as a tool, but as the “be-all and end-all” of the creative process.
“I just feel it’s a shame we need to do it,” Dunbar said.
Slop cleanup jobs may not last. It’s unclear whether cleaning up AI-generated slop will stay a reliable source of work.
Dunbar isn’t convinced it will. She thinks much of the work to humanize AI could disappear within five to ten years as models improve and let businesses generate usable content on their own.
McConnell sees it differently, arguing that AI still needs to be “babysat” to catch mistakes an untrained eye might miss, from logos that break brand rules to objects that pop in and out of videos. “We need people [to fix] problems that AI is creating,” he said.
But Van Linda is already carving out a different path. After he stoSince he stopped taking AI cleanup jobs late last year, he’s seen more clients—some of whom were unhappy with AI—looking for original, hand-drawn work. “Honestly, this kind of thing has become really distasteful to me,” he said about AI cleanup work.
Lisa, a freelance graphic designer, updated her profile to say she no longer works with AI-generated images, but she still gets requests to edit them. She occasionally takes on cleanup jobs from clients who come to her after getting poor results with AI. However, some clients “straight up lie” and try to pass off the images as their own original work. “They say ‘I don’t use AI,’ but there’s a [Google] Gemini logo at the bottom,” she said.
But just in case AI makes her job obsolete, she has a backup plan. Lisa has started returning to the canvases, watercolors, and other physical art she practiced before pursuing a career in graphic design. She predicts that within “another year or two,” AI will be able to handle logo and packaging design as well as she can, leaving only “the best of the best” able to make a living. “I don’t know if I’m good enough to survive the digital stuff,” she said.
Frequently Asked Questions
Here is a list of FAQs based on the scenario of freelancers cleaning up AIgenerated content
General Background
1 What exactly is soulless AIgenerated content
Its text that is technically correct but lacks a human touch It often reads as repetitive generic and overly formal with no unique voice personal anecdotes or emotional depth It feels like it was written by a robot following a formula
2 Why are freelancers suddenly so overwhelmed with this kind of work
Many businesses have rushed to adopt AI tools to cut costs and speed up content production They generate a high volume of drafts quickly but then realize the quality isnt good enough to publish They hire freelancers to fix the drafts which creates a new massive wave of lowpaying tedious cleanup work
3 What does the cleanup process actually involve
It involves rewriting entire paragraphs to sound more natural factchecking claims the AI hallucinated adding a consistent brand voice improving the flow removing repetitive phrases and injecting realworld examples or personal insights that AI cant create
4 Why are freelancers so unhappy about this trend
It feels like a step backward Instead of creating original work theyre spending hours polishing turds Its often underpaid for the time it takes creatively draining and feels like they are doing the AIs job for it rather than using their own expertise
The Freelancers Perspective Problems
5 Is cleaning up AI content actually harder than writing from scratch
Often yes It can be harder to rewrite a bad AI draft than to start with a blank page You have to fight against the structure and tone the AI created whereas starting fresh gives you full control It also requires extra vigilance to catch subtle factual errors
6 Why is the pay usually so bad for this type of work
Clients often assume that because the AI did the heavy lifting of generating the draft the editors job is quick and easy They dont factor in the time it takes to factcheck restructure and humanize the text leading to lowball offers
7 What are the biggest red flags to look for in an AIgenerated draft
Look for excessive use of the word delve generic transitions like In conclusion a
