What happened
CNN reports that LinkedIn has expanded efforts to identify and limit low-quality AI-generated posts. The platform added a flagging button in August, and LinkedIn said users clicked it more than 1 million times in its first two weeks. The company is now using AI systems trained with help from former journalists and social media experts to identify content it considers generic or inauthentic.
CNN reports that LinkedIn previously offered tools to edit or rewrite posts with AI but recently scaled those offerings back. The company has also created a team described as largely consisting of former journalists and social media experts, who help train LinkedIn’s AI model on what the company considers authentic and useful content.
According to CNN, when the model flags a post as low-quality and AI-generated, the post becomes less likely to be distributed beyond the author’s direct connections. LinkedIn has also used a creator-management team to advise executives and other users to write with more personal detail and imperfection rather than relying on polished, generic text.
The platform’s August flagging lets users indicate that a post “seems like AI slop.” LinkedIn said the button was clicked more than 1 million times in its first two weeks. CNN also cites a July Pangram report that found more than 40% of LinkedIn long-form posts were AI-generated, but that measurement is attributed to Pangram and is not independently established here.
Why it matters
The change shows a major professional platform treating AI-generated content quality as a distribution and retention issue, rather than only as a disclosure or spam problem. If LinkedIn reduces the reach of generic AI-written posts, professionals and companies may need to add more personal judgment, editing, and firsthand experience to content intended for broad visibility. The available evidence is limited to LinkedIn's statements reported by CNN.
For users who depend on LinkedIn for professional visibility, the reported ranking change could make AI-assisted publishing less predictable. A post can be grammatically polished yet receive less reach if LinkedIn’s systems judge it to be generic or insufficiently personal.
The policy also raises a practical distinction between using AI as an editing aid and allowing it to produce an entire post. CNN reports that LinkedIn is steering users toward adding their own observations, experience, and voice, but the source does not establish how the platform technically distinguishes those cases.
Because LinkedIn is owned by Microsoft, the strategy reflects a tension between promoting workplace AI tools and limiting the low-quality content those tools can help produce. CNN reports that LinkedIn is not opposed to AI generally; its stated focus is the content’s usefulness and authenticity.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
crm_get_transaction(id='4092').Which component of an AI application is the machine-learning model itself?
What to watch next
LinkedIn's detection criteria, review process, appeal options, and treatment of edited or partially AI-assisted posts remain unclear. It is also unknown whether the system is broadly available across markets and account types, how false positives are handled, and whether LinkedIn will publish independent evidence supporting its claimed reduction in AI-slop views.
LinkedIn has not publicly detailed the model architecture, training data, detection thresholds, or categories of content that trigger reduced distribution, according to the material provided.
The source does not state whether users are notified when their reach is reduced, whether they can challenge a , or whether the policy applies equally to text posts, videos, newsletters, and other formats.
The reported performance improvement comes from LinkedIn itself. Independent testing would be needed to assess whether the claimed 40% reduction reflects less low-quality content, reduced visibility caused by ranking changes, changes in user behavior, or another factor.