LinkedIn Cracks Down on AI-Generated Spam With User Flags and Updated Classifiers

What LinkedIn Actually Changed

The platform introduced a “Seems like AI slop” option accessible through the three-dot menu on any post. Members can use it to flag content they believe was generated or heavily automated with minimal human input.

LinkedIn Chief Product Officer Hari Srinivasan has stated that combining these user reports with updated detection classifiers has reduced views of content the platform classifies as AI slop by 40% in recent weeks. That’s a meaningful reduction in reach, even if it doesn’t yet tell us whether the volume of AI-generated posts being published is actually declining.

Alongside the flagging tool, LinkedIn has also:

• Updated its automated classifiers for identifying AI-generated or AI-assisted content
• Removed a feature that offered to enhance posts with AI
• Begun rolling out author notifications when their posts receive AI-related flags from other members

The author notice is framed as feedback rather than accusation—a careful distinction, given that AI-assisted writing is widespread and not inherently problematic.

The Scale of the Problem

The concern isn’t theoretical. Data from Pangram, an AI-detection company, offers useful context on how prevalent fully AI-generated content has become across major platforms.

Among LinkedIn longform posts (250+ words) reviewed between April and June 2026, 41% were flagged as fully AI-generated by Pangram’s classifier. For shortform posts (50–250 words), the figure was 30%. Both numbers are notably higher than comparable figures on X/Twitter, Reddit, or Substack.

This positions LinkedIn as the platform with the highest apparent concentration of AI-generated content among those measured—which helps explain why the user response to the flagging tool has been so immediate and large.

What LinkedIn Is Targeting—and What It Isn’t

LinkedIn appears to be drawing a deliberate line. The focus is not on AI-assisted writing in general—drafting, editing, restructuring—but on content that shows signs of mass production with little human involvement: repetitive posts, generic professional advice, automated comments designed to inflate engagement metrics.

This distinction matters for anyone who uses AI tools as part of their writing workflow. The platform is not signaling a ban on AI assistance. It is signaling lower tolerance for content that looks like it was produced and distributed at scale with no meaningful human layer on top.

The updated classifiers extend an effort LinkedIn began earlier in the year, when it announced action against spammy AI-generated comments. The latest changes bring that same logic into the main feed.

What the 40% Figure Actually Means

The 40% reduction in views is worth reading carefully. It measures how widely posts already classified as AI slop are being distributed—not how many such posts are being published. LinkedIn’s classifiers are doing less amplification work for flagged content, but the publishing behavior of users hasn’t necessarily changed.

That gap matters. If the reduced reach doesn’t feed back into changed posting behavior, the platform will be managing distribution rather than solving the underlying incentive to post AI-generated content at volume.

Practical Takeaway

If you use AI tools to help draft LinkedIn posts, the immediate risk is low—provided the output reflects a genuine point of view and isn’t indistinguishable from templated content. The practical threshold LinkedIn appears to be enforcing is not “did AI touch this?” but “does this look like it was produced at scale with no human intent behind it?”

For teams managing LinkedIn presence at volume, that’s a meaningful distinction worth building into your content process now, before classifier sensitivity increases further.