Freelancers Are Drowning in AI Slop Cleanup Work, and Now They're Walking Away
A graphic designer in Spain spent most of 2024 fixing broken AI-generated logos for low pay. Her story captures a quiet crisis spreading across the freelance economy.

Key points
- By 2025, 90% of one freelance designer's logo and packaging requests involved fixing AI-generated work, not creating original designs.
- That AI cleanup work made up 60% to 70% of her annual income.
- Clients consistently offered less money for cleanup jobs than for original design work.
- Some AI-generated files were so badly made that she had to rebuild them entirely from scratch.
- By late 2025, she stopped taking the work altogether.
Lisa is a freelance graphic designer based in Spain. Before 2022, her work was straightforward: clients hired her to create logos, packaging, brand identities. Then ChatGPT arrived, and with it, image-generating AI tools that let almost anyone produce a passable-looking design in seconds.
What happened next surprised her.
The jobs did not dry up. They mutated.
As first reported by The Guardian AI, Lisa began receiving a flood of requests to fix AI-generated designs rather than make new ones. Fuzzy images that could not be printed at full size. Logos built from corrupt or unusable digital files. AI-generated artwork that looked acceptable on a phone screen but fell apart under professional scrutiny. Clients had used an AI tool, got something almost-good-enough, and then needed a human to cross the finish line.
So what exactly is 'AI slop cleanup'?
Think of AI image generators as fast but clumsy apprentices. They can produce something that looks like a finished design in seconds, but the underlying file is often a mess: blurry edges, wrong colour formats, missing layers, or shapes a printing press cannot read. A trained designer then has to open the file and fix, or sometimes completely rebuild, what the AI produced.
That is the work Lisa found herself buried in.
By 2025, nine out of every ten logo or packaging requests she received were cleanup jobs. That work paid 60% to 70% of her annual income, which sounds like job security until you factor in the conditions. Clients who had already paid an AI tool almost nothing expected human expertise at a matching price. Some files were so broken that fixing them took longer than designing from scratch would have. And Lisa worried about a thornier issue: copyright. AI image generators are trained on other people's artwork, often without permission, and designs they spit out can carry legal risk that passes silently to whoever finishes the file.
Why did she stop taking the work?
Exhaustion, mostly. Low pay, unclear legal ground, and the creeping feeling that she was propping up a process she did not believe in.
By the end of 2025, Lisa began turning the cleanup jobs down.
Her story is not unique. Across illustration, copywriting, translation and photography, freelancers are reporting the same pivot: the work has not vanished, it has shifted. Instead of paying for original human skill, clients want humans to polish the rough edges off machine output, at machine-level prices.
For anyone hiring a freelancer right now, that tension is worth understanding. AI tools can cut your upfront cost. But the hidden cost, paid either in a freelancer's time or in a printed logo that smears and warps, does not disappear. It moves.
Lisa's answer was to stop subsidising it.



