LinkedIn tightens action on inauthentic content, DSA data shows

LinkedIn removed 70.5 million pieces of spam and inauthentic content in the second half of 2024, according to data cited by Social Media Today from LinkedIn's Digital Services Act transparency reports. That figure is not a record to be proud of. It is a measure of how badly the platform's feed has been colonised by automated content farms, and how much engineering effort is now being redirected from distribution features to content hygiene.

The DSA reports cover LinkedIn's EU user base, which held flat over the reporting period. Flat user growth in a market where professional networking is still expanding is its own signal. When a platform stops growing in a major geography and simultaneously escalates content-removal operations, the two facts belong in the same sentence. The feed quality problem is almost certainly part of the stagnation story.

The mechanism behind the removal surge

LinkedIn's enforcement scaled not because the rules changed, but because the volume of AI-generated filler accelerated faster than organic posting did. Text generators made it trivially cheap to publish at scale: keyword-stuffed motivational posts, scraped industry statistics dressed up as original insight, profile-stuffing sequences designed to manufacture connection requests. The platform's classifiers had to catch up. They are still catching up.

The practical consequence for legitimate publishers is a stricter distribution environment. When a platform is under DSA scrutiny and processing tens of millions of removals per reporting cycle, it tunes its filters conservatively. Posts that pattern-match against inauthentic signals, even if they are entirely genuine, face a higher probability of suppressed reach. Those signals include: posting at high frequency with low engagement variance, using near-identical sentence structures across posts, and generating connection velocity that looks automated.

This is the collateral effect that most commentary on the DSA numbers misses. The story is not simply that LinkedIn is cleaning up its act. It is that the cleanup imposes compliance costs on authentic voices who happen to resemble, statistically, the bad actors.

What this means for institutional publishers

For a communications director at a multilateral institution, an industrial group's chief executive, or a financial services firm managing several executive profiles, the implication is concrete. LinkedIn's classifiers are increasingly treating content behaviour, not just content, as the authenticity signal. A posting programme that runs like a content factory, even one staffed by humans, will be read as one.

The variables that now carry more weight are qualitative ones: specific comments that reflect genuine familiarity with a topic, profile visits that follow a piece of long-form content, saves and shares from accounts with their own credible posting history. A senior partner at a professional services firm who posts twice a week and receives twelve substantive comments from named peers in their field is, from the algorithm's perspective, credibly human. The same firm's LinkedIn page posting daily from a templated content calendar is increasingly a candidate for suppression, regardless of intent.