LinkedIn published an update on 15 September 2026 on what it calls AI slop. The numbers in it are the ones that traveled: more than 1 million members have used the Seems like AI slop button since it launched in August, and views of content LinkedIn classifies as slop are down 40%.
The part that matters to anyone with a media budget is not in that announcement at all. It is in a help page written for advertisers, and it says something the announcement does not.
Write a weak post and you find out. Run a weak ad and you never do.
Here is what the announcement promises creators:
“If a post receives enough community feedback, the author may see a message in their post analytics to help them understand how their content is being received.”
That is a feedback loop. Write something the room finds hollow, and the room's verdict reaches you.
Now the advertiser help page, "AI slop feedback option for ads on the LinkedIn feed." Members can flag paid content too. LinkedIn says members can share feedback by selecting Seems like AI slop “on member posts, comments, and ads shown on the LinkedIn feed.”
And then:
“Please note, at this time, we won't notify you if a member selects this option for your ad, and we also don't provide this information as a reporting metric.”
No notification. No column in Campaign Manager. No export. The signal exists, members are pressing it, and the one party paying for the impression is the one party who cannot see it. LinkedIn is direct about where it goes:
“we're currently only using this information for our internal learning purposes to help us improve our products so we can provide the best member experience.”
Two limits worth knowing before anyone panics
It is feed ads only. The help page states the option “isn't available for ads shown on the right rail of LinkedIn or on the LinkedIn Audience Network.” If your program is mostly right rail or Audience Network inventory, this does not touch it yet.
LinkedIn is not against AI-written ads. This is worth quoting in full, because the coverage of LinkedIn's slop policy keeps turning into "the platform is hunting AI content," and that is not what the page says:
“LinkedIn supports the responsible use of AI to draft and improve advertising content. We focus on whether content provides value, not only on how it was created.”
Value, not method. A thoughtful ad drafted with AI is fine. A hollow ad written by a person is not.
The caveat that keeps this honest
LinkedIn has not said that a flagged ad loses delivery, reach or impressions. The 40% drop belongs to organic content that LinkedIn's own classifiers catch, and no equivalent figure has been published for ads. What the help page describes is aggregate learning, not a documented penalty.
So the accurate way to hold this: a quality signal you cannot see, not a penalty you can measure. That is still worth acting on, because the direction of travel across every LinkedIn statement this year has been the same. In May, LinkedIn said generic content is less likely to spread beyond your immediate network. In July, it said comments that repeat what the post already said get shown less prominently. The slop button is the member-facing edge of the same push.
LinkedIn wrote down what a bad ad is
Buried under a heading called "What can make your ad feel low value," LinkedIn lists exactly what it means. Your ad can feel like slop, in its words, if it:
“Uses generic claims that could apply to any business.”
“Repeats ideas without adding useful information.”
“Lacks a clear brand perspective or point of view.”
“Doesn't provide specific information about your product, service, or offer.”
“Uses template-style language or images that aren't relevant to the intended audience.”
Read that as a checklist for your live ads, written by the platform that decides what your budget buys. Most ad copy fails at least two of those lines, and the failures are usually fixable in one pass.
What each line looks like in practice
Generic claims that could apply to any business. "We help teams work smarter" describes every software company ever funded. Swap the claim for the thing only you can say: which team, doing what, and what changed.
Repeats ideas without adding useful information. The second and third sentence of most B2B ads restate the first in new words. Cut them and the ad gets better, not shorter in substance.
Lacks a clear brand perspective. An ad with no point of view has nothing for a reader to agree or disagree with. A stated opinion is what makes a stranger stop, and it is the one thing a model cannot invent for you.
No specific information about the product, service or offer. If a reader cannot tell what they would be buying, the ad is an impression, not an advertisement.
Template-style language or images not relevant to the audience. Stock abstractions and interchangeable phrasing are precisely the pattern members are now being asked to flag.
Content Credentials are a different thing
LinkedIn has also adopted the C2PA provenance standard, which it calls Content Credentials. When an image or video carries those credentials, LinkedIn shows an icon, and a member can click it to see whether AI was used to make or edit the asset. It is easy to conflate that icon with the slop flag. LinkedIn separates them explicitly: “The presence of a Content Credentials icon doesn't mean that LinkedIn or its members consider the content AI slop.”
Provenance is about how something was made. The flag is about whether it was worth the reader's attention. Only the second one is a judgment.
What to do this week
Audit five ads against the five lines. Take your highest-spending live creatives and read each one against LinkedIn's list. Rewrite the ones that fail. Name the customer, name the number, cut the adjectives.
Brief your writers and your tools with those five lines. If AI drafts your ad copy, the five criteria are a better brief than any tone instruction, and they come from the platform itself.
Watch what you can still measure. You will not see flags, so watch the things that move when creative goes stale: click-through decay within a frequency band, comment sentiment, the gap between impressions and any engagement at all.
Keep the human point of view in the copy. LinkedIn's stated test is value, not authorship. A specific opinion, a real number, a named outcome. Those are the things a member does not report.
Your audience already has a button for everything else.
Verified 16 September 2026 against LinkedIn's newsroom post of 15 September 2026 and the LinkedIn Marketing Solutions help article "AI slop feedback option for ads on the LinkedIn feed." Quotations are reproduced exactly. LinkedIn has published no delivery penalty for flagged ads; the 40% figure describes organic content classified by LinkedIn's systems.
Working out where AI actually fits in your marketing?
I write these while building the systems behind them: measurement, creative pipelines, and agents that do real work. Connect on LinkedIn and tell me what you are working on. That is where these conversations start.
Connect on LinkedIn