Unchecked AI output doesn’t remove work. It moves the work onto whoever reads it, and your payroll never sees the transfer. We think AI slop grenades are the most expensive habit a small team can pick up in 2026, and the fix is a rule, not a tool. Whoever hits send owns every word.
The short version
- Shopify’s CEO says staff toss “slop grenades” at each other, meaning AI output sent without checking.
- BetterUp’s survey found 40% of US desk workers got something like it in a month, and each case took about two hours to fix.
- Don’t ban the tools and don’t buy a better model. Put one sentence in your working agreement saying the sender answers for anything they send, whoever or whatever wrote it.
In a September interview, Shopify CEO Tobias Lütke described the habit in two words. Someone generates a long AI-written message, fires it at a colleague, and never checks whether it’s right. The colleague then has to read it, fix it, or feed it to another model to shrink it. He asked why we invented “decompression and recompression.” Put plainly, it’s a hidden transfer of labour, and most small firms are paying for it without knowing. Lütke used the term on The Knowledge Project, per Business Insider’s report of 17 September.
What is a slop grenade, and why does it cost so much?
A slop grenade is AI-generated work passed to a colleague without the sender checking it. The sender saves time up front. The receiver can lose far more, because checking someone else’s work is slower than checking your own, and the receiver can’t tell which parts are wrong without reading all of it.
BetterUp’s survey gave the pattern a second name, workslop, and some numbers. BetterUp Labs and the Stanford Social Media Lab surveyed 1,150 US desk workers. 40% said they had received workslop in the previous month, and each case took about 1 hour 56 minutes to resolve. 53% admitted sending it at least some of the time. Both shares are self-reported, so treat them as a direction and not a measurement.
What could it cost a 50-person firm?
Use BetterUp’s own inputs and the answer is uncomfortable. Suppose 40% of your 50 people, so 20, get one piece of workslop in a month. At roughly two hours each, that’s 40 hours, a full working week of one person’s time, spent undoing other people’s shortcuts. BetterUp puts the dollar figure at about $186 per affected person per month, a self-reported estimate, which would be $3,720 a month for those 20.
That’s our arithmetic applied to a survey of US desk workers. Your number may be a third of that or double it. The trouble is that it never shows up as a line item. It shows up as everyone being busy and nothing moving.
Why don’t the usual fixes work?
Buying a better model doesn’t help, because the problem isn’t the model’s accuracy. Nobody is accountable for reading the output before it travels. A banned-tools policy fails too. People route around it, and you lose the real gains along with the junk.
Lütke’s own line points at the answer. “Humans take responsibility,” he said. The machine can inform you better, but a person still has to put their name on the result. Shopify, for what it’s worth, isn’t an AI sceptic. He says its internal agent handles a large share of its code changes, possibly as much as half. The rule is about who answers for the output, not whether to use it.
The rule that works
Put one sentence in your team’s working agreement: if you send it, you’ve read it, and you answer for every claim in it. Then back it with three habits.
- Anything over a screen long carries a two-line summary written by the sender, not the model.
- Numbers, names and dates get a source or a “checked by” mark before they leave your desk.
- Receivers are allowed, even expected, to bounce unreviewed work back with no hard feelings.
We covered the wider version of this problem in our piece on the hidden time tax of AI, and our comparison of the big three assistants is useful if you’re deciding what to standardize on first.
Where this could be wrong
Nobody has measured workslop in a real firm like yours, and BetterUp’s figures come from self-reports, with the dollar value an extrapolation. Lütke’s remarks are an anecdote from one very large, very AI-heavy company. They give a real habit a good name, but they don’t prove how common it is at 30 people.
We’d change our view if a month of counting rework in your own team showed no rise since AI use took off. Then the rule is cheap insurance, not a fix.
The sceptic’s best case
The sceptic says people always sent sloppy emails, and AI has just given the old problem a new label. That’s partly right. Bad memos predate chatbots.
What changed is volume and polish. An AI draft looks finished, so the reader trusts it longer before spotting the gap. A sloppy human email announces itself, and a slick wrong one doesn’t, which is why the old etiquette won’t cover it.
What to watch
- Whether your own team’s rework goes up as AI use does. Ask managers to count it for a month.
- Whether vendors start shipping “review before send” prompts in their assistants.
- Any independent study of workslop in firms under 500 staff.
Frequently asked questions
What is a slop grenade?
It’s Shopify CEO Tobias Lütke’s phrase for AI-generated material that someone sends to a colleague without checking it, so the colleague has to review, fix or summarize it.
What is workslop?
It’s BetterUp Labs and the Stanford Social Media Lab’s term for AI-generated work that looks finished but lacks the substance to be usable. In their survey, 40% of US desk workers said they got some in a month.
How can a small business stop AI slop at work?
Make the sender responsible for everything they send. Require a human-written summary on long items and a source on every number, and let receivers return unchecked work.
Written by Harper Singh, an AI editorial persona at AI Magazine Canada. This is analysis and opinion. The cost example is our own arithmetic on a published survey, not a measurement of any firm. Written and fact-checked on 8 October 2026. Sources are linked on the claims they support.
