Universal Ads TV launch partner. Read more →
    Back to Blog
    Paid Social

    LinkedIn Ranks Comments by Relevance After Finding 30% Were AI-Generated: What It Means for Paid Social Trust

    Adlyse Team
    October 4, 2026
    7 min read
    LinkedIn Ranks Comments by Relevance After Finding 30% Were AI-Generated: What It Means for Paid Social Trust

    Higher ROAS. 10x less work.

    Put Adlyse on your accounts. Daily audits, agentic optimization, and reports your clients actually read.

    Request a demo

    LinkedIn changed how comments display under posts, moving from a shared, chronological order to a personalized ranking based on each individual viewer's professional interests, connections, and engagement activity, a shift reported August 9, 2026. It lands in the middle of a monthslong effort LinkedIn has been public about since at least July 2026: a "Seems like AI slop" reporting button, the removal of its own AI post-enhancement tool, and a chief product officer stating the platform blocks hundreds of thousands of automated spam comments daily and billions of broader automation attempts over a couple of months. This isn't one announcement; it's an escalating pattern worth reading as a trend, not a single fresh finding, and it's relevant to any Executive thinking about brand safety on a platform where AI-generated content is now a measured, acknowledged problem rather than a hypothetical one.

    Key facts
    • LinkedIn's shift to relevance-based, per-viewer comment ranking was reported August 9, 2026; it replaces a shared chronological order with personalized ranking.

    • A Pangram Labs analysis of roughly 57,000 public LinkedIn posts found about 30% of comments posted between April and June 2026 were entirely AI-generated; this is a third-party estimate without a published detection methodology or stated error rate, not an official LinkedIn figure.

    • Separately, LinkedIn's own chief product officer stated in July 2026 that the platform blocks hundreds of thousands of automated spam comments daily and has prevented billions of other automation attempts over the preceding couple of months.

    • LinkedIn introduced a "Seems like AI slop" reporting button and removed its own "enhance your post" AI writing tool in favor of a proofreading feature, both part of the same broader crackdown.

    • Available source material does not establish a direct, confirmed causal link between the 30% AI-comment estimate and the comment-ranking change specifically; this post treats them as related parts of the same broader trend, not as cause and effect.

    Is the 30% figure a LinkedIn-confirmed statistic, or an outside estimate?

    An outside estimate, and that distinction matters enough to state plainly. Pangram Labs, a third party, analyzed roughly 57,000 public LinkedIn posts and estimated that about 30% of comments posted between April and June 2026 were entirely AI-generated. The underlying report doesn't publish a detection methodology or an error rate, which is a real limitation: a figure like this, without a disclosed method for distinguishing AI-generated text from human text, should be treated as a directional estimate from one research group, not an audited platform-wide statistic LinkedIn itself has confirmed.

    That caveat doesn't make the broader trend less real. LinkedIn's own chief product officer, speaking separately and on the record, described blocking hundreds of thousands of automated spam comments daily and billions of other automation attempts over roughly the preceding two months, which is LinkedIn's own acknowledgment that automated, low-quality content at scale is a problem it's actively fighting, independent of whether the specific 30% figure holds up to scrutiny.

    Why did LinkedIn change how comments are ranked, specifically?

    The reported change moves from a shared chronological order (everyone sees comments in the same sequence) to a personalized, relevance-based order (each viewer sees comments ranked by their own professional interests, connections, and engagement activity). Available reporting describes LinkedIn as simultaneously fighting engagement pods and AI spam in comment sections, but doesn't draw an explicit, confirmed line connecting that fight to this specific ranking mechanism as its direct cause. The more defensible read: this is one of several changes LinkedIn is making around the same time, in the same general direction (surfacing what's actually relevant and trustworthy to a given viewer, rather than what simply arrived first or was posted at scale), not a single fix targeted at the AI-comment problem specifically.

    What's the broader crackdown this sits inside?

    A pattern that's been building publicly since at least July 2026. LinkedIn added a "Seems like AI slop" button that lets any user flag content they believe is low-quality or AI-generated. It removed its own "enhance your post" AI writing assistant, replacing it with a proofreading tool explicitly framed as preserving a user's authentic voice rather than generating content for them, a notable reversal for a platform that had previously been pushing its own AI writing tools. And LinkedIn's chief product officer has been direct in public statements that "AI slop is a top priority for all of us," describing the daily and monthly scale of automated content LinkedIn says it's blocking.

    Read together, the comment-ranking change, the reporting button, the AI-writer rollback, and the public statements describe a platform treating AI-generated, low-authenticity content as a trust problem it needs to visibly address, not a side issue. Whether any single mechanism (ranking, reporting, tool removal) is solving the underlying problem is a separate, unresolved question from whether LinkedIn is treating it seriously, which the pattern of actions suggests it is.

    Why does this matter for an Executive thinking about paid social, not just organic LinkedIn use?

    Because brand safety and AI-content governance questions that used to be abstract, "what happens if AI-generated content floods a platform we spend money on," now have a concrete, acknowledged case study. If a platform you're running paid campaigns on is publicly fighting a measured AI-content authenticity problem in its own organic layer, two practical questions follow directly: does that authenticity problem bleed into the environment your paid content appears next to (comment sections under your own sponsored posts, for instance), and does the platform's response to it (ranking changes, reporting tools) change how your organic and paid presence is actually seen by the audiences you're trying to reach.

    Neither question has a settled answer yet based on available material, and this post doesn't claim one. What it does claim is that "AI slop at scale" has moved from an industry talking point to a platform-acknowledged, partially-quantified problem on a channel many B2B and professional-audience advertisers treat as core, which makes it a reasonable line item in a brand-safety or governance review this quarter, not next year's.

    What should an Executive actually do with this, right now?

    Ask your social or brand team whether comment sections under your own paid or organic LinkedIn content have shown signs of automated or low-quality engagement. This is a direct, checkable question against your own account, more useful than reasoning abstractly about industry-wide figures.

    Treat the 30% figure as a directional signal, not a number to cite as fact in your own reporting. It's a third-party estimate without disclosed methodology; citing it as if LinkedIn confirmed it risks repeating an unverified number as settled.

    Watch whether this pattern extends to other platforms you advertise on. LinkedIn is the platform with the most public, quantified response right now, but AI-generated low-quality content at scale isn't a LinkedIn-specific problem; it's a reasonable question to ask about any platform where your brand's paid content sits next to user-generated comments.

    Frequently asked questions

    No. That figure comes from a third-party analysis by Pangram Labs of roughly 57,000 public posts, covering comments from April through June 2026. The underlying report doesn't disclose a detection methodology or error rate, so it should be read as a directional estimate, not an official LinkedIn statistic.

    Reported August 9, 2026: LinkedIn moved from a shared, chronological comment order to a personalized, relevance-based order, ranking comments differently for each viewer based on their professional interests, connections, and engagement activity.

    Not confirmed directly. Available reporting describes LinkedIn as fighting both engagement pods and AI spam around the same period, but doesn't establish the ranking change as a targeted fix for the AI-comment estimate specifically. They're related parts of the same broader trend, not confirmed cause and effect.

    LinkedIn introduced a "Seems like AI slop" reporting button, removed its own AI post-enhancement writing tool in favor of an authenticity-preserving proofreading feature, and its chief product officer has stated publicly that the platform blocks hundreds of thousands of automated spam comments daily and billions of broader automation attempts over recent months.

    It raises the question seriously. Whether AI-generated, low-authenticity content in comment sections affects how paid content on the platform is perceived isn't settled in available material, but it's a reasonable addition to a brand-safety or governance review given how publicly LinkedIn itself is treating the issue.

    Related reading

    If platform-trust questions like this one need a documented place in your governance review, Explore Adlyse →