On 30 July, LinkedIn started limiting how far generic AI posts travel. They are not deleted. They stay visible to the people who already follow you, and they stop spreading past that.

Three things changed at once.

A report button. Every post now carries a “seems like AI slop” option in its menu. Anyone can press it. It hides the post from that reader, and it feeds LinkedIn’s classifiers, which is the real purpose: the reports are training data.

Less reach for what those classifiers catch. Flagged content appears less in recommendations outside your own network. Accounts flagged often enough are told so privately, in their analytics.

The rewrite tool is gone. This is the easiest change to miss and the hardest to argue with. LinkedIn removed “enhance your post”, the feature that rewrote your draft to sound better, and replaced it with one that proofreads and leaves your voice alone.

What it means

Your feed should get quieter. Pangram Labs scanned about a million posts and found 41% of long LinkedIn posts, 250 words or more, were fully AI-generated. That was the highest share of any platform they measured, and their CEO called it a lower bound. Most of it says nothing. Cutting its reach is good for everyone still writing something real.

You can still use AI. LinkedIn said so plainly, and it kept its own AI features. What loses reach is empty content, not machine-written content. A post drafted with AI that makes a real point does fine. A post you typed yourself explaining that social media is powerful for business does not, and never should have.

That distinction, between content written by a machine and content that says nothing, is the one we built the product around.

Why we score content authority

Every tool can write a post. Almost none of them tell you whether the post was worth writing. That gap is the entire reason slop exists, and it is why Lightr scores a draft before you publish rather than reporting on it weeks after.

We score three things.

Originality. Could anyone in your industry have published this word for word? If yes, there is nothing in it that is yours, and nothing for a reader or a classifier to tell apart from the rest of the feed.

Depth. Does the reader learn anything they could not have guessed from the headline? Specifics carry a post, because specifics are the part that cannot be generated from a topic alone.

Clarity. Can you state the argument in one sentence? If you cannot, neither can your reader.

You see the score and the reasoning while you can still act on it: publish it, fix it, or drop it. The three questions are written out in full in Score before you publish, and you can run them by hand on anything, no tool required.

The rest of the design

Scoring catches weak drafts. Two earlier choices stop them being produced.

Lightr writes in your voice instead of smoothing every business toward the same neutral, confident, faintly breathless register. That register is precisely what LinkedIn just removed from its own product.

And it starts from your material rather than a blank page: your positioning, your past posts, what you actually do. Content anchored in real specifics reads differently because it is different.

Why this holds

None of it is built to beat a detector. Hari Srinivasan, LinkedIn’s chief product officer, was unusually candid about why the button exists at all: “Slop is hard to define and the definition changes; this lets us tune our models and make better feeds.” Anything tuned to this month’s classifier is built to break.

We scored for originality when nothing was forcing us to, because a post nobody remembers was already a bad outcome. LinkedIn now charges you reach for the same thing.

Write things worth reading. The cost of ignoring that used to be invisible. Now it is your audience.