As artificial intelligence has made it easy and inexpensive to create huge volumes of content, the internet is flooded with this 'junk.' Platforms like YouTube, Pinterest, and Spotify have begun to limit the amount of spammy AI-generated content that users see.
However, LinkedIn has been particularly affected. When browsing this professional network earlier this year, users might have encountered a stream of bulleted lists decorated with emojis, links to 'B2B sales,' and requests to follow or comment for more information.
According to a report published in July by the research firm Pangram, which tracks AI content online, over 40 percent of long posts on the platform were created using AI. This is significantly higher than the approximately 30 percent on X and Medium, and also higher than the smaller percentage on Reddit and Substack.
In response, LinkedIn is intensifying its fight against such low-quality content. Dana Roth, LinkedIn's chief editor who heads the platform's content strategy, told CNN: 'I know you think you are doing the right thing with your perfectly structured list and call to action at the beginning, and you followed all the advice from LinkedIn consultants online. But what works is what is human.'
There is a certain irony since LinkedIn belongs to Microsoft. This tech giant actively promotes the use of its AI tools for everything from writing text to coding, yet it faces stiff competition from companies like OpenAI and Anthropic. Previously, LinkedIn offered AI tools for editing or rewriting posts, although it has recently reduced these offerings.
Nevertheless, low-quality AI-generated posts pose a business problem for LinkedIn. As Roth noted, they all start sounding the same. If users see too many generic AI-generated posts instead of content that truly teaches something or provides deeper insight into professional connections, LinkedIn risks losing them to other platforms.
In August, LinkedIn launched a button allowing users to flag posts that 'look like AI junk.' Within two weeks of the launch, this feature was clicked over a million times. The company is now trying to use its own AI systems to filter out this 'junk' and remind LinkedIn users how to post in a human style.
The company has formed a team consisting mainly of former journalists and social media experts who can recognize generic, sloppy content. These specialists help train LinkedIn's AI model on what authentic and valuable publications look like. If the model flags a post as low-quality AI-generated content, the likelihood of it spreading beyond the user's direct contacts decreases.
A key element of the strategy is also the author management team, which acts as 'content therapists' for executives, subject matter experts, and other LinkedIn users. Many of them are accustomed to the convenience of using AI to create outwardly polished posts, as Roth observed. He said that users often run text through Claude or Gemini or other tools, publish it on LinkedIn, and it performs poorly. They then approach the team with the question: 'Why did this perform badly?' The best solution, according to him, is to return the content to them and ask: 'Did you enjoy reading this? Do you see yourself in this?'
Instead, the team encourages users to embrace imperfection in their posts—for example, filming a video by simply leaning a phone against a coffee mug instead of shooting in an expensive studio, or quickly writing a post to start a dialogue even if it is not a complete thought.
These efforts show early signs of success. The platform reported last month that improvements in LinkedIn's detection systems led to a 40 percent decrease in views of content classified as 'AI junk.' Roth emphasized that this does not mean LinkedIn is completely against AI, adding that 'not everyone is born a natural communicator.' However, if content is not edited or personalized, it is less likely to reach a wide audience on the platform.
'We are just trying to push people toward more humanity,' he concluded. 'That is what it takes to cut through the noise.'