News · AI news

LinkedIn is demoting AI-generated comments. What should agencies change?

LinkedIn has turned its effort against "AI slop" from posts to comments. In a post on his own profile, reported by Social Media Today on 15 November 2026, LinkedIn's Chief Product Officer Hari Srinivasan described two changes: new classifiers that reduce low-quality comments in the default "most relevant" view, and smarter notifications for creators.

What changed

The classifiers assess how relevant a comment is to the post it sits under, and suspected AI-generated comments are treated as less valuable. They are not deleted. They are ranked lower, so anyone reading a thread in the default view should see fewer of them.

The notification change works on the other side of the conversation. According to Srinivasan, creators will be alerted to "high-quality comments you're likely to care about", which means a generic comment is less likely to reach the person it was written to impress.

LinkedIn has given one number. Srinivasan said: "In testing we're already seeing a >20% reduction in views of slop comments, and they will be rolled out to everyone soon." No rollout date has been given, and LinkedIn has not said how the classifiers decide what counts as AI-generated or how often they get it wrong.

What it means for agencies

Comments are one of the easiest forms of engagement to hand to a tool: a short reply, drafted against someone else's post, published under a founder's or a client's name. On LinkedIn that tactic now has a stated cost: the comment is less likely to be seen by readers, and less likely to be surfaced to the creator.

A few practical consequences:

  • Audit any automated or semi-automated commenting you run on client profiles. If the output reads as interchangeable, assume it is now being ranked down.
  • Fewer, specific comments are worth more than many generic ones. A comment that adds a fact, a disagreement or an example from the client's own work is the kind the relevance classifier is described as rewarding.
  • Report this to clients plainly. If a leadership profile's comment activity was part of an engagement target, the target may need to change to something like replies earned or conversations started.
  • Do not try to make AI comments harder to detect. The signal LinkedIn describes is relevance, and the durable fix is a comment that is actually relevant.

It is worth keeping the scope in view. This is a ranking change to comments on one platform, announced in a single post, with one testing figure and no published methodology. It does not say AI-assisted writing is banned, and it does not say anything about posts. What the classifier is described as targeting is low relevance, not the tool that wrote the words. AI has simply made low-relevance comments cheap to produce at volume.

For agencies, the conclusion is the same one that applies to posts. A person who knows the client should read and stand behind anything published in the client's name, including the comments.

Your brand, impossible to scroll past.

Brandaris is the AI social studio agencies run on — listening, creation, human approval, publishing and analytics in one workflow. Nothing ships without a human yes.

Apply for early access