LinkedIn Killed "Plus in Comments" and AI Content: How the Algorithm Works Now

LinkedIn has rolled out an algorithm update called Authenticity Update. The platform has started fighting template AI content and primitive schemes like “Put a plus in the comments — I’ll send you a guide”. Such posts now get their reach cut, and accounts can go into shadow ban without warning. 

What Changed

LinkedIn introduced a new metric — Depth Score. It measures not the number of likes, but the quality of engagement: how long the user read the post, whether they clicked “See more”, saved the post, or forwarded it in private messages. If a post gets 100 likes in a few minutes but is scrolled past in the first second — the algorithm understands that the engagement is fake and reduces reach. 

What gets blocked: 

• AI texts. The algorithm detects the standard structure of language models: uniform sentence length, template lists, lack of real emotions and specific terms. 

• External links. Posts with links to third-party sites lose up to 60% of reach. LinkedIn wants to keep users inside the platform.

• Manipulation posts. Posts like “Write the word ‘want’ in the comments and I’ll send you a guide” are now officially considered spam. 

• Mutual engagement groups. Groups where participants like and comment on each other’s posts are detected and blocked.

What Works in 2026

PDF carousels and documents. This format gets 2–3 times more time spent than regular text or images. The algorithm sees this and promotes such posts.

Native video up to 30 seconds. Videos uploaded directly to LinkedIn get additional reach. Be sure to add subtitles — most users watch videos without sound.

Longreads with expertise. Posts of 1200–1800 characters with real cases, numbers, and professional slang perform better than short motivational quotes.

LinkedIn Newsletter. Newsletters bypass the feed algorithm — subscribers receive notifications directly. This is the only way to guaranteed reach your audience.

What Arbitrageurs Should Do

Old schemes with mass posting via API and auto-commenting no longer work. Accounts get blocked, and networks are detected within a couple of days.

Working approaches:

• Quality over quantity. Better 5–7 accounts with detailed history and expert content than 100 empty profiles.

• Remove AI markers. Use neural networks for drafting structure, but rewrite the final text manually. 

• Hide links. The “plus in comments” mechanic no longer works. Drive traffic to pre-landers through native mentions in the text or use trusted accounts to seed in the first comment.

• Use PDF carousels. Content embedded in presentations is perceived differently by the algorithm — this gives additional organic reach.

LinkedIn’s audience remains one of the most solvent in B2B, but the rules have changed and you’ll have to adapt to them. Now those who build expert content win, not those who stamp out AI fillers.