Do you use ChatGPT or Claude to prepare your LinkedIn posts?
Then the problem is no longer simply whether your text "sounds human".
LinkedIn is actually beginning to reduce the distribution of content that it considers to be a sloping AI, i.e. generic, repetitive, automated or devoid of any real point of view.
And now your audience can also participate in the signal.
More than a million people have already used the option Seems like AI slopAccording to Hari Srinivasan, Chief Product Officer of LinkedIn. At the same time, LinkedIn indicates that its members see about 40 % less content classified as AI slop that a few weeks ago.
Should we stop using AI to publish?
No.
LinkedIn explicitly states that theIA can help write. What he seeks to limit is the substanceless content that gives the impression that nobody really had something to say before pressing "Generate".
For consultants, coaches, freelancers, executives and creators B2B, the consequence is much more interesting than a simple "IA penalty":
your personal experience and your perspective become algorithmic benefits again.
LinkedIn isn’t fighting AI: it’s fighting content without substance
It is the most important shade of all this news.
LinkedIn defines the problem as IA content generic, unworked and without real perspective or substance.
In an official May publication 2026, LinkedIn already explained:
- that it remained acceptable to use AI to help write
- that publications must nevertheless represent the voice and real point of view of their author
- platform systems learn to differentiate content providing expertise, context and perspective of repetitive or generic content.
In other words:
IA used AI slop.
The problem arises when artificial intelligence replaces almost entirely:
- the idea
- the experience
- the reasoning
- opinion
- example
- proof.
A concrete example
Let's take this post:
"Artificial intelligence is transforming the world of work. Companies need to adapt quickly. Here are five tips for successful digital transformation. »
The text can be grammatically perfect.
Who's talking?
What experience lies behind it?
What did you observe that others don't know?
Why should your audience read your publication rather than the other fifty saying exactly the same thing?
Compare now:
" Last week I tested the automation of three administrative tasks in an SME. Two really saved time. The third created more work than she did. That's why. »
The subject remains artificial intelligence.
The difference is huge.
There are now:
one experience + one tension + one observation + one promise.
This is what transforms text into content.
Over a million people have already reported ‘AI slop’
LinkedIn recently added an option to allow members to report content that seems to be in this category.
It appears in the menu associated with the publications.
According to Hari Srinivasan, more than one million people had already used this option shortly after its launch.
It's huge. But we must avoid misinterpretation.
An alert does not automatically destroy your scope
LinkedIn explicitly states thatno individual feedback decides on the distribution of a publication.
The platform uses several signals and claims to have put in place mechanisms to prevent users from unfairly targeting someone by massively reporting it.
So it's not:
a user clicks → your post disappears.
Community feedback becomes one signal among others.
This nuance is particularly important, because recognizing automatically that a text has been assisted by an AI is far from trivial.
LinkedIn announces about 40 % fewer views of content classified as AI slop
This is probably the most important figure for creators.
Hari Srinivasan explains that, overall, LinkedIn members now see about 40 % less content than LinkedIn classifies as AI slop a few weeks ago.
Be careful the meaning of the number. That does not mean:
“Every post written with ChatGPT loses 40 % of its reach.”
That would be wrong.
LinkedIn talks about the amount of content classified as AI slop now seen by members.
The platform had also indicated as early as May that this type of publication was less likely to be distributed beyond the author's immediate network.
It is this second information that must really interest creators.
The real risk concerns discovery
Imagine you have 5 000 relationships and subscribers. Your publication may affect part of your existing network.
But the important growth on LinkedIn also comes from the ability to be shown to people who do not follow you yet.
If LinkedIn considers your content as generic, it is this expanded distribution that can become more difficult.
To develop his personal branding, it's a real problem.
LinkedIn even starts to warn authors
The platform now introduces another step.
When a publication receives enough Community feedback, LinkedIn can display in Post Analytics a message informing the author that some members believe that its publication resembles IA-generated content.
The stated objective is more educational than punitive. LinkedIn wants to help the author understand how his content is perceived.
It's an interesting approach. Because the real problem isn't always:
" Was this publication actually written with an AI? »
It may be:
" Does this publication look so much like all the automatically generated texts that readers think it is? »
And this second problem exists even if you wrote each word yourself.
You can write yourself... and still produce
This is where many creators will be wrong.
They're gonna think:
"No problem, I don't use ChatGPT. »
But generic content and AI content are not synonymous.
A human can write perfectly:
"Never abandon your dreams. Every failure is a lesson. Success requires work, discipline and perseverance. »
No IA required.
And yet, there is almost no distinctive information.
Conversely, a professional can use Claude to structure an analysis from fifteen years of experience and produce something particularly useful.
The question is therefore not:
Who typed on the keyboard?
The question is:
where does the thought come from?
The 6 signs that can give your publication an "AI slop" aspect
There is no official checklist to predict LinkedIn ranking.
But several features are clearly incompatible with the direction expressed by the platform.
1. You could replace your name with any other
Review your publication. If it could be published exactly as it is by:
- hR consultant
- a business coach
- a marketing agency
- a SaaS
- a student;
without anyone noticing the difference, your editorial identity is too weak.
2. You have no proof
You say:
"The AIA makes it possible to significantly increase productivity. »
What is the basis for this statement?
A test?
A study?
An experiment?
A client?
A source?
Without proof or context, you simply repeat a popular idea.
3. Your post uses an automatic structure
For example:
" It's not X.
That's Y."
Then:
"Here 7 lessons that I would have liked to know earlier. »
Then a succession of very short and predictable sentences. No structure is bad in itself.
The problem arises when structure becomes more identifiable than your thought.
4. You publish a lot without increasing the depth
If you are alone and publish five major analyses a day on different topics, your audience may naturally wonder how this production is possible.
LinkedIn also monitors automated or very large-scale publishing behaviour. Its documentation specifies that an automated or inauthentic activity may lead to a limitation of visibility or even an account restriction.
5. Your comments simply repeat the post
LinkedIn explicitly cites as a problem the answers that merely reformulate the original content without bringing anything new.
The famous comment:
"That's right! IA is actually transforming our way of working. Thank you for this very relevant sharing. »
probably helps no one.
6. There is no personal experience
You don't have to tell every publication about your life.
But strong professional content often contains at least one element difficult to generate without you:
- one observation
- a decision
- one error
- one capture
- internal data
- one test
- a reasoned opinion
- an anonymized real case
- a contradiction with the dominant idea.
How to use ChatGPT or Claude without losing your voice
The solution is not to go back to 2019] and ban AI in your workflow.
Use it differently.
Wrong method
Prompt:
"Write me a viral LinkedIn post about the importance of personal branding. »
You ask the AI:
- the subject
- the idea
- the corner
- the arguments
- examples
- writing.
She's doing almost everything. You only provide the theme.
Best method
Start by writing:
"Yesterday, I analyzed ten profiles of consultants. Seven were unable to explain clearly what they were selling in their first three lines. The problem was not their expertise: it was their positioning. Here are the three mistakes I've spotted... »
Then give your notes to the AI and ask:
"Help me structure these observations in a clear post. Keep my examples and my perspective. Let me know the generics instead of inventing new ones. »
IIA becomes editor.
No ghost author. This is an essential change.
The VIFLY method : Material → AI → judgement → publication
A good use of artificial intelligence can take four steps.
Step 1: Human Material
Always start with something from you:
- vocal note
- observation
- call client
- experience
- analysis
- opinion
- capture
- numbers
- problem encountered.
Step 2: IA Support
Then use the AI to:
- structure
- clarify
- challenger
- condenser
- identify objections
- find the passages weak.
Step 3: Human Judgment
Read it again.
Remove what you would never have said. Add the nuances. Check the facts.
Strengthen examples.
Step 4:
Before you publish, ask this question:
"What part of this content could only come from me? »
If you don't have an answer, rework it.
This is also precisely the approach developed in the VIFLY guide IA and creators: creating without becoming generic : the AI must increase an original material rather than replace the point of view of the creator.
Your LinkedIn profile must now prove that you are a real source
This fight against generic content is not just about posts. It also reinforces the value of a coherent profile.
LinkedIn explains that its new Feed system uses language models that can better understand what a publication really talks about and how this topic corresponds to the professional interests of members.
This implies an interesting consequence.
The more coherent your topics, expertise and positioning, the more understandable your professional world is.
For example:
Profile A
Entrepreneur
Profile B
SEO consultant for French e-commerce. Google Search analysis, content and conversion to help stores increase their skilled traffic.
The second gives immediately more context.
Your bio, content and offers should tell the same story.
Do not seek to "appear human": be precise
From the moment a platform fights against AI content, a new industry of absurd advice usually appears:
Make mistakes voluntarily.
Use more emojis.
Avoid words.
Ask ChatGPT to write "more human".
It's a bad direction.
Human content is not defined by spelling error. It is defined by the presence of a thought located.
Compare:
" It is essential to adapt its marketing strategy to new consumer expectations. »
With:
"On the 22 reservations received by this coach in July, 17 came from the same Reel yet published three weeks earlier. So it was not the frequency that created the conversion: it was the accuracy of the subject. »
The second example is human because it contains a particular reality.
Not because he uses a particular vocabulary.
LinkedIn could even create an opportunity for real experts
There is another way to read this news.
If LinkedIn effectively reduces the volume of generic content in its recommendations, space available for truly distinctive content increases.
Imagine a feed containing:
100 publications.
While 30 are automatic variations of the same five ideas, they occupy much of the available attention.
Reducing their distribution can give space to:
- niche expertise
- less known professionals
- field returns
- original views
- technical analyses.
For small audiences, this is potentially excellent news.
You don't need to be able to produce more than AI.
You must produce something that she cannot invent for you:
your real experience.
LinkedIn content must also lead somewhere
Another problem remains intact.
You can write an excellent human publication, get 50 000 impressions and do not take away any opportunities.
Why?
Because visibility and conversion are two different problems.
The course should be:
visibility → understanding → trust → action → opportunity.
LinkedIn mainly manages the start of the course.
It's up to you to organize the next one.
LinkHub can become your feed outlet
A VIFLY LinkHub in particular, it is possible to centralise:
- your social media
- your important content
- your resources
- your offers
- your contact details
- your booking.
It also makes it possible to distinguish certain sources of traffic, which can help to understand whether LinkedIn actually generates qualified visitors rather than just impressions.
For a LinkedIn consultant, the objective could be:
Post → profile → LinkHub → audit → booking.
Content attracts. Your infrastructure converts.
For coaches and consultants, connect the expertise to a simple action
Suppose you published an excellent analysis on:
"The three reasons why commercial teams use ChatGPT poorly. »
A leader finds you. He read several publications. He clicks on your profile. He wants to talk to you.
If he must now:
- search your site
- find your form
- wait for your answer
- exchange four emails to find a date
you add friction when the intent is maximum.
For appointment-based activities, VIFLY Booking can be used as natural continuity of content: the expertise attracts the demand, the Booking simplifies the action.
Of course, you don't have to turn every post into a commercial pitch.
But your profile must allow an interested person to know what to do next.
Use VIFLY IA to improve your thinking rather than make volume
The best use of AI for LinkedIn is probably no longer:
"Give me 30 posts for the month. »
It becomes:
"This is my experience. What interesting conclusions have I not yet made? »
Or:
"Challenge my opinion as a skeptical client would. »
Or:
" What passages in my text resemble generalities that anyone could write? »
Or alternatively:
" What assertions require a source? »
It is this usage logic that VIFLY AI can accompany: learn how to use generative tools as think-tanks rather than as machines to automatically produce an online presence.
What if LinkedIn tells you that your post looks like the AI?
Do not immediately "beat the detector".
Start by rereading the publication as a human.
Check the central idea
Can you summarize it in an interesting sentence?
If the answer is:
" You have to be authentic on LinkedIn",
probably not enough.
Look for what really comes from you
Did you:
- a story?
- experience?
- proof?
- an example?
- a number?
- an opinion?
Check your structure
Does the text seem artificially cut?
Does every sentence look like a punchline?
Is the introduction more promising that the content will bring?
Check the volume
Do you publish so much that every idea becomes superficial?
Analyze performance
Do not just focus on impressions.
Look:
- comments actually developed
- profile views
- new relevant subscribers
- clicks
- conversations
- incoming requests.
A post of 5 000 impressions that generates two prospects can have more value than a post of 100 000 views producing only likes.
What to stop doing now
The four practices I would remove as a priority are simple.
First reflex: stop the totally open prompts of the type "write me a viral post".
Second reflex: stop automated comments that do not add anything to the discussion. LinkedIn explicitly targets automation and low human involvement responses.
Third reflex: stop publishing only to keep a calendar.
A bad publication does not become better because it was scheduled Tuesday at 8 h.
Fourth reflex: stop optimizing to "look human".
Optimize for have something to say.
Conclusion: LinkedIn does not bury the AI, it increases the price of generic content
LinkedIn's new battle against the "AI slop" could easily be summed up by an anxious title:
« LinkedIn penalizes ChatGPT. »
Yet it would be missing the most interesting.
LinkedIn itself uses artificial intelligence massively in its recommendation system. Its new generation Feed relies on LLMs to better understand content and professional interests.
The problem is therefore clearly not the existence of AI. The problem is what it makes possible on a large scale:
produce a lot of content without much thought.
And in this environment, scarcity changes. Yesterday, what was rare was the ability to produce.
Today, producing text is almost free. What becomes rare again is:
experience, judgment, evidence, originality and perspective.
That's great news for real experts.
Your next step: audit your 10 latest LinkedIn posts
Take your last ten publications and give them one point for each of these items:
- personal experience
- concrete example
- identifiable perspective
- verifiable information
- really useful lesson
- next clear action.
Don't look for a magic algorithmic score.
Look for the pattern.
If your publications have beautiful hooks but almost no clean material, do not change your ChatGPT prompt: change what you give ChatGPT.
Start tomorrow your next content not with:
"What should I publish? »
but with:
"What did I observe, learn, test or understand this week that my audience doesn't know yet? »
Then only use the AI to help you better express.
In 2026, being assisted by the IA is not the problem. Be interchangeable east.
FAQ
Does LinkedIn penalize all content written with ChatGPT?
No. LinkedIn explicitly states that it is acceptable to use theIA to help write. The platform focuses on reducing the dissemination of generic, repetitive, automated or non-real content.
What is "AI slop" on LinkedIn?
LinkedIn uses this term to refer to low-effort AI content that may seem clean but lacks perspective, expertise or substance.
Does LinkedIn actually reduce the scope of AI slop?
LinkedIn's Chief Product Officer indicates that members generally see 40 % less content than the platform ranks as AI slop than a few weeks ago. This number does not mean that each IA post automatically loses 40 % scope.
Can a user lower my range by signaling me?
Not by himself. LinkedIn states that no individual feedback determines distribution and that several signals are used with abuse protection.
Can I automate my LinkedIn comments?
LinkedIn prohibits unauthorized third-party tools that automate messages, comments, likes, publications or other activities. Their use may lead to restrictions.
How to use the AI without producing generic content?
Start with your own material experience, notes, data, opinion or real case and then use theIA to structure, challenge or edit. Avoid simultaneously delegating the idea, angle, arguments and examples.
Sources
LinkedIn Hari Srinivasan, Chief Product Officer updated to "AI Slop", published around 21 August 2026. Primary source of the figure of more than one million feedbacks, of the overall decrease of about 40 % of exposure to AI slop classified content, of protections against abusive reporting and of the arrival of feedback in Post Analytics.
LinkedIn , 20 may 2026. Official documentation explaining that the AI can help write but that LinkedIn reduces the dissemination of generic content without point of view or substance. LinkedIn also indicated that its first systems correctly identified generic content in 94 % cases during internal testing.
LinkedIn Engineering , 12 march 2026. Official source describing LLM-based Feed, semantic understanding of publications, and multiple signals used to customize recommendations.
LinkedIn Help Automated activity / Prohibited software. Official documentation recalling that third-party tools automating publications, comments, likes, messages or other unauthorized activities may violate LinkedIn rules and lead to limitations or restrictions.
Business Insider 25 August 2026, . Recent secondary source confirming and contextualizing Hari Srinivasan's statements and the evolution of Post Analytics.