Workslop is AI-generated work that looks complete but does not give the recipient enough to act. Researchers at Stanford’s Social Media Lab and BetterUp coined the term in September 2025to describe this problem.
How common is workslop, and what does it cost?
The original survey covered 1,150 full-time US desk workers. Four in ten had received workslop from a colleague in the previous month. Recipients estimated spending nearly two hours repairing each incident on average. The study asked people to recall that time; it did not time their work.
What it does to trust
Forty-two percent trusted the sender less, half saw the sender as less capable, and 32% said they would rather not work with that person again. These are reported reactions, not observed career outcomes.
Who is producing it, and do they know?
SHRM’s 2026 workplace research drew on more than 5,000 US workers. Forty-one percent used AI at work; among those users, 44% said some of their own output qualified as AI slop. At least some users recognize the problem in their own work. The survey does not tell us why they still send it.
What later research found
Workday’s January 2026 study surveyed 3,200 active AI users at large employers in November 2025. It found that 85% reported saving one to seven hours a week with AI, while rework absorbed nearly 40% of those reported savings. Only 14% consistently achieved what the study called clear positive net outcomes. These estimates cover workplace tasks, not writing alone.
How to avoid producing workslop
Before you send, ask what the recipient can do with the document. Name the decision, owner, facts, and open questions. Then check every claim, cut generic prose, and read the result aloud. Zero Slop packages that review as a free, open-source skill. The fuller argument is in The Slop Tax.