it costs time
Two hours, every time.
Each workslop message takes the receiver nearly two hours to decode, repair, or redo. Workers estimate 15.4% of everything they receive is now slop.
AI now writes the messages your company sends — to clients, prospects, and the people who decide your reputation. Most of it is confident, polished, and quietly hollow. This is the case for protecting the one thing AI can’t fake: trust.
AI use at work has doubled — yet MIT found 95% of organizations see no measurable return on it. One reason hides in plain sight. BetterUp Labs and Stanford’s Social Media Lab named it: AI-generated content that masquerades as good work, but lacks the substance to actually move a task forward. It’s polished. It’s confident. And it quietly hands the real work to whoever receives it.
Each workslop message takes the receiver nearly two hours to decode, repair, or redo. Workers estimate 15.4% of everything they receive is now slop.
An “invisible tax” of roughly $186 per employee per month — over $9 million a year for a 10,000-person firm. None of it shows up on a line item.
After getting slop, about half saw the sender as less capable and reliable; 42% as less trustworthy; and 32% didn’t want to work with them again.
None of these are hypothetical. They’re the messages already landing in inboxes — the AI “tells” that make a real person look careless.
“Hi — I came across your impressive profile and saw you’re a leader in your space. I thought our solution would be perfect for someone like you.”
The tell: personalization with zero specifics. The recipient knows instantly a machine wrote it, and that no human spent thirty seconds on them.
“Our best-in-class, game-changing platform delivers unmatched, world-class results and guarantees ROI within the first quarter.”
The tell: every adjective, no evidence. In a regulated firm, “guarantees” alone is a compliance incident.
“Peer firms are seeing massive results with this approach — you’re falling behind. Worth a quick call to walk through the numbers?”
The tell: a number-shaped claim with no number behind it. Manufactured urgency the sender can’t actually support.
“Per our policy, that’s not something we cover — you should have reviewed the disclosure. There’s nothing further we can do here.”
The tell: fluent, fast, and wrong — or needlessly cold. To the customer it isn’t “the AI.” It’s your firm.
The same generic, confident, hollow text is turning up in every channel where a person used to write something real — at work, up the org chart, and even on a first date.
“Dear Hiring Manager, I am writing to express my profound enthusiasm for this exciting opportunity. As a results-driven, detail-oriented team player, I am passionate about leveraging synergies to drive impactful outcomes.”
The tell: every cover letter and LinkedIn DM now reads identically. The AI didn’t make the candidate stand out — it made them invisible. Recruiters skim past in seconds.
“Team, as we navigate this dynamic landscape, we remain committed to leveraging our core strengths to unlock value and drive sustainable growth across all verticals going forward.”
The tell: three paragraphs, zero information. Everyone scrolls looking for “so… are there layoffs or not?” Leadership spent its credibility to say nothing.
“Circling back — we’ve made significant progress on key workstreams and are aligned on next steps to ensure successful execution moving forward.”
The tell: your teammate still has to ask “…so is it actually done?” The update created a meeting instead of replacing one.
“Hey! Your profile really caught my eye — you seem like an adventurous soul with a great sense of humor. I’d love to get to know the amazing person behind the photos. 😊”
The tell: sent to 200 people, word for word. The one human who might’ve mattered can tell instantly that nobody read a thing about them.
Slop isn’t a one-time irritation. It taxes time, buries the messages that actually matter, and quietly wears people down.
Every slop message pushes the real effort onto whoever receives it — about two hours per incident to decode, correct, or redo. The time the sender “saved” was only borrowed from everyone after them: $9M+ a year per 10,000 people.
When everything sounds polished and says nothing, people stop reading closely. The message that mattered — the real client question, the urgent flag, the genuine note — gets skimmed past and missed in a feed of confident filler. Reply rates fall with it.
Being the human who vets the machine’s plausible-but-hollow output, all day, is its own exhausting job: decision fatigue, eroded trust between teammates, and the quiet burnout of never knowing whether what you’re reading was actually thought through.
The research measured slop between coworkers. The moment that same draft is addressed to a client, a prospect, or a regulator, the math changes — because the safety net disappears.
When client conversations moved to channels firms didn’t supervise — personal texts, WhatsApp — the SEC and CFTC wrote more than $2 billion in penalties across roughly 100 firms, $50M+ each at names like LPL, Raymond James, Ameriprise, and Edward Jones. Off-channel was about where messages went. AI is about what writes them. The firms that paid for the WhatsApp problem will not want to get caught twice.SEC & CFTC off-channel / recordkeeping enforcement sweep, 2022–2024.
Most firms reach for the same three things: a policy PDF, an annual training, and tool monitoring. But awareness training teaches employees what AI is — not how to use it safely on a live client message — and the rest of the stack is blind to both halves of the problem.
Slop is built to pass a glance. It looks done. By design, nobody flags it until the receiver tries to use it — and by then it has already left. A policy nobody opens at the moment of send catches none of it.
57% of employees hide that they use AI at all; 69% report a stigma around admitting it. Your people are already drafting client messages with AI — you just don’t know which ones. A Stanford review found 77% of the hardest AI-adoption problems had nothing to do with the technology.
Answering shadow AI with heavier monitoring deepens the distrust that drives people to hide in the first place. The fix isn’t policing the person. As the researchers put it, the mere presence of AI is no guarantee of a productivity gain — what’s missing is control at the point where the work goes out.
We’re not here to sell you an answer. We’re here to name the problem out loud and ask the question every team is about to face.
Imagine the slop never left the building. Every AI-assisted message that went out had already been checked — not the person policed, the message itself — so the honest ones flew and only the risky ones ever paused.
People use AI openly instead of hiding it. Clients stop getting confident nonsense. The messages that matter actually land. And the trust that’s quietly leaking away stops leaking.
What if control lived at the moment of send — not in a policy nobody opens? What if admitting you used AI were safe, because the system caught the risk instead of policing the person?
Routine messages flow untouched. Only the risky ones slow down — just long enough for the right person to look. And every decision leaves a record someone could stand behind.
We don’t think this is unsolvable. We think it’s the next thing worth building — and the conversation worth starting now, before the regulators start it for us. What would you want it to look like?
Outbound governance isn’t a someday idea. There’s a working answer today — checks at the moment of send that catch the risky message before it ships, without policing the people doing honest work.