AI can turn a few sentences into pages of prose. Some employers have had enough.

Employers person a caller connection for workers utilizing AI to write: Stop the slop.

Fed up pinch bloated and confusing content, from task proposals to position updates, bosses are instituting policies that outline wherever reasonable soul AI use ends, and overreliance begins. They opportunity simply urging workers to curb the behaviour hasn't been enough.

A general argumentation creates a shared "touchpoint" alternatively than requiring leaders to many times explicate what is expected, said Clare Jones, CEO of the travel-app shaper Polarsteps.

The Amsterdam-based company, founded successful 2015, released a group of principles earlier this twelvemonth connected erstwhile and really its astir 100 labor should usage AI for soul writing. (Most of the consumer-facing transcript successful its app is produced by master in-house writers, Jones said.)

Polarsteps' expectations include, for example, disclosing erstwhile unit person utilized AI and ensuring the last transcript reflects their ain reasoning and voice. Jones said labor tin still usage AI to brainstorm and draft, but they are expected to use their ain captious reasoning earlier sharing the result.

How slop spread

Driving the crackdown has been a gradual summation successful time-wasting slop successful the years since OpenAI kicked disconnected the generative AI gyration pinch the launch of ChatGPT successful precocious 2022.

A precocious 2025 study from BetterUp and Stanford's Social Media Lab recovered that 40% of 1,150 full-time US table workers had received AI slop from a coworker successful the preceding month. Respondents said dealing pinch each lawsuit took astir 2 hours connected average.

"The captious wide is location now," said Jessica Zwaan, vice president of group strategy and operations astatine the AI-powered HR package institution Leapsome. "Everyone is conscionable producing ample amounts of writing."

At a caller soul meeting, she said she spent much clip reference the prep materials than she did successful the gathering itself. "I don't cognize if I've ever wanted a clear condemnation longer," she said. "Longer is not better."

Zwaan theorizes that workers successful wide are misusing AI because they've been told it is an imperative, and whitethorn interest that sending little notes alternatively of lengthy AI-generated documents will make them look lazy.

"A batch of group subordinate productivity pinch volume, and actually, productivity is clarity," she said.

Leapsome's AI penning argumentation was rolled retired successful mid-August and instructs its 130 labor successful Berlin and New York to guidelines down what they say, attraction connected clarity, and beryllium mindful of the length.

"If a short punctual produced a agelong document, it's not worthy reference for the aforesaid logic it wasn't worthy writing," it says.

Blaming AI for ambiguous write-ups is besides discouraged.

"'AI wrote it' isn't an reply if personification asks what you meant," the line states. "Every connection should beryllium thing you back."

'Just disregard that'

AI slop has go truthful ubiquitous that techies person coined a nickname for perpetrators: "meat proxies." The word was popularized by German package Niklas Gruhn successful an August 3 blog station and refers to those who stock unchecked AI outputs.

A general argumentation tin thief minimize that move by making it little awkward for workers to situation colleagues' AI-generated writing, said Sophie Alpert, a lead technologist astatine Clay, a sales and trading automation platform. She wrote the company's policy, which was first shared pinch its astir 100-person engineering squad successful mid-June earlier being rolled retired to each 550 of its labor successful early August.

"I wanted to create abstraction for group to push backmost without it seeming for illustration a individual judgment," she said.

The caller rules astatine Clay require labor to guidelines down each thought and condemnation successful their documents and to make their writing worthy reading.

"Any other clip that readers request to walk to understand what you meant incurs a multiplicative costs connected the team's time," it says. "Conversely, if you walk other clip to make your archive clear and concise earlier sending it, you are paying a one-time costs that each scholar will use from."

Alpert said she was inspired to return action aft colleagues sent her 5 overtly AI-generated documents successful 1 week. When she asked them to explicate immoderate confusing parts, she said she was told, "just disregard that."

In a motion of really broadly Alpert's consequence resonates, a LinkedIn station astir it by Clay's CEO, Varun Anand, went viral past month. It had astir 10,000 reactions and much than 700 reposts arsenic of September 3.

Guardrails, not bans

The AI penning policies astatine Polarsteps, Leapsome, and Clay don't magnitude to an outright AI ban. Instead, the standards are aimed astatine preventing workers from passing disconnected unexamined AI output arsenic their ain reasoning — aliases burdening colleagues pinch it.

So far, bosses down them opportunity the reactions they've seen from workers person been mostly positive.

"I deliberation a batch of group were like, 'Thank goodness someone's doing thing astir this,'" said Zwaan, adding that immoderate labor asked the institution to spell further by restricting AI successful employee-to-employee messages and marketing copy.

At Polarsteps, the caller guidance is already having a affirmative effect.

"We're getting truthful overmuch quality contented now, which is really lovely," said Jones.