Employers have a new message for workers using AI for writing: stop the doldrums.
Bosses are fed up with bloated and confusing content, from project proposals to status updates, and are instituting policies that indicate where reasonable internal AI use ends, and overconfidence begins. They say simply urging employees to curb the behavior hasn’t been enough.
A formal policy creates a shared “touchpoint” rather than leaders having to repeatedly explain what is expected, says Clare Jones, CEO of travel app maker Polarsteps.
The Amsterdam-based company, founded in 2015, released a set of principles earlier this year on when and how its approximately 100 employees should use AI for internal writing. (The majority of the consumer-facing copy in the app is produced by professional in-house writers, Jones said.)
For example, Polarsteps expectations include disclosing when employees have used AI and ensuring the final copy reflects their own reasoning and voice. Jones said employees can still use AI to brainstorm and draft concepts, but they will be expected to apply their own critical thinking before sharing the result.
How the sloppiness spread
The driving force behind the crackdown has been a gradual increase in time-wasting doldrums in the years since OpenAI kick-started the generative AI revolution with the launch of ChatGPT end of 2022.
A late 2025 survey from BetterUp and Stanford’s Social Media Lab found that 40% of 1,150 full-time US desk workers had received AI slop from a colleague in the past month. Respondents said each case took an average of almost two hours to resolve.
“The critical mass is there now,” says Jessica Zwaan, vice president of people strategy and operations at AI-powered HR software company Leapsome. “Everyone is just producing large amounts of writing.”
At a recent internal meeting, she said she spent more time reading the preliminary material than she did during the meeting itself. “I don’t know if I’ve ever wanted a clear sentence any longer,” she said. “Longer is not better.”
Zwaan theorizes that employees generally are misuse of AI because they’ve been told it’s absolutely necessary, and they may worry that sending short notes instead of long, AI-generated documents will make them look lazy.
“Many people associate productivity with volume, and really productivity is clarity,” she said.
Leapsome’s AI writing policy was rolled out in mid-August and instructs its 130 employees in Berlin and New York to stand behind what they say, focus on clarity and be mindful of length.
“If a short prompt produces a long document, it’s not worth reading, for the same reason it wasn’t worth writing,” it says.
It is also not recommended to blame AI for ambiguous articles.
“‘AI wrote it’ is not an answer when someone asks what you meant,” the guidance says. “Every word should be something you stand behind.”
‘Just ignore that’
AI slop has become so ubiquitous that techies have coined a nickname for perpetrators: “proxies for meatThe term was popularized by German software developer Niklas Gruhn in an August 3 blog post and refers to those who share uncontrolled AI outputs.
A formal policy can help minimize that dynamic by making it less difficult for employees to question colleagues’ AI-generated writing, says Sophie Alpert, a principal engineer at Clay, a sales and marketing automation platform. She wrote the company’s policy, which was first shared with the approximately 100-strong engineering team in mid-June, before being rolled out to all 550 employees in early August.
“I wanted to create space for people to push back without it seeming like a personal judgment,” she said.
The new rules at Clay require employees to stand behind every idea and sentence in their documents and make their ideas come true writing that is worth reading.
“Any additional time readers have to spend to understand what you meant comes at a multiplying cost to the team’s time,” it says. “Conversely, if you spend extra time making your document clear and concise before sending it, you pay a one-time cost that every reader will benefit from.”
Alpert said she was inspired to take action after colleagues sent her five openly AI-generated documents in one week. When she asked them to explain some confusing parts, she said she was told to “ignore that.”
To show how widely Alpert’s response resonates, a LinkedIn post about it by Clay co-founder Varun Anand went viral last month. It had nearly 10,000 comments and more than 700 reposts as of September 3.
Guardrails, no bans
The AI writing policy at Polarsteps, Leapsome and Clay does not amount to an outright AI ban. Instead, the standards are intended to prevent employees from passing off unexamined AI output as their own thinking – or burdening colleagues with it.
So far, the bosses behind them say the response they’ve seen from employees has been largely positive.
“I think a lot of people said, ‘Thank God someone is doing something about this,’” Zwaan says, adding that some employees have asked the company to go further by limiting AI in employee-to-employee messages and marketing copy.
At Polarsteps, the new guidance is already having a positive effect.
“We’re getting so much human content now, which is really wonderful,” Jones said.
Correction: September 6, 2026 — An earlier version of this story misstated Varun Anand’s role. He is Clay’s co-founder.
