LinkedIn chief product officer Hari Srinivasan says the company blocked “billions of other automation attempts (posting at scale, slop) in the last couple months alone” as it cracks down on low-quality AI content. On Thursday, LinkedIn backed that enforcement with a new user-facing control: a “Seems like AI slop” button that appears when you click the options button on a post.

The point is simple, and it’s also tactical. Users can flag content they believe is low quality or overreliant on AI, and LinkedIn says the goal is to “tune our models and make better feeds.” In his post, Srinivasan also acknowledged the slipperiness of the label: “Slop is hard to define and the definition changes.” Translation: LinkedIn expects the concept of “AI slop” to evolve, and it wants ongoing feedback loops from humans to improve how it ranks content.

That matters because LinkedIn is not just another social app. It is a professional network where discovery, recruiting, business development, and brand building all depend on feed quality. When AI-written content floods platforms, the harm is not only spammy posts. It is dilution of signal. If your feed becomes a river of repetitive, low-effort automation, everything from job searches to partner outreach gets noisier, and trust erodes. LinkedIn’s response is therefore two-pronged: increase machine enforcement (bot detection) and add a lightweight human reporting option that can help refine what the machines treat as “slop.”

The enforcement headline is the big one. Srinivasan said LinkedIn detects and blocks “hundreds of thousands of automated slop comment attempts every day.” Pair that with the “billions” number for other automation attempts in the last couple months, and you get the scale of what LinkedIn is battling: not single bad actors, but systematized content farming that can run at speed. This also gives decision-makers a clue about where LinkedIn is investing. It is not relying solely on user reporting or after-the-fact moderation. It is building automated detection pipelines strong enough to filter at volume, then using the new “AI slop” button to keep those pipelines aligned with what users consider unacceptable.

There is also a product-level shift behind the scenes. LinkedIn is removing its “enhance your post” tool and replacing it with a feature that proofreads posts without changing a user’s voice, while reinforcing profile and page verifications. That is an important nuance for executives watching this space: LinkedIn is not banning AI use wholesale. Srinivasan explicitly clarified that using AI is not always bad. He said many people refine their thoughts with AI, and the new “AI slop” tool would let users know when their post sounds “inauthentic.” In other words, the company is trying to separate assistance from impersonation. Assistance can help people write better. Slop, in LinkedIn’s framing, is what pushes content toward automation, repetition, and inauthentic presentation.

This move lands in a broader market context that has been accelerating since ChatGPT’s release in 2022. Fortune notes a study by AI text detection tool Pangram found 40% of long-form posts and 30% of short form posts on LinkedIn were flagged as fully AI-generated. Whether any given detector is perfect is a separate debate, but the business reality is clear: platforms are facing an arms race where generative tools get cheaper, easier, and more capable, while audiences increasingly demand authenticity and relevance. For LinkedIn, feed ranking is the battleground, because that is where engagement happens and where the platform either concentrates trustworthy signal or amplifies noise.

The competitive landscape is also telling. Snapchat announced that it would no longer recommend fully AI-generated videos via its Spotlight feature, while carving out exceptions for videos edited or enhanced using Snapchat’s own AI creative tools, complete with transparency indicators. Meta, by contrast, is leaning into generative content by developing its own family of generative media models. Earlier this month, Facebook and Instagram parent Meta launched Muse Image, which can create AI-generated images based on references like people, objects, clothing, styles, and environments, and it also previewed Muse Video. Meta says images include an invisible watermarking system called “Content Seal” designed to identify them even if they are cropped, resized, or screenshotted, and it plans to extend that system to Muse Video soon. During Meta’s second-quarter earnings call, CEO Mark Zuckerberg said Muse Image and Muse Video would “dramatically expand the universe of content that people can discover across our platform,” adding that it would make services “more useful and engaging.”

What LinkedIn is doing sits between those extremes. It is not going full ban like Snapchat’s Spotlight recommendation change, and it is not full embrace like Meta’s generative discovery push. Instead, LinkedIn is building moderation and feedback mechanisms that aim to keep AI assistance from turning into AI sludge. For boards, investors, and platform leaders, the second-order implication is that “AI content policy” is turning into “AI distribution policy.” The question is not only what gets posted, but what gets surfaced, recommended, and rewarded. If LinkedIn’s feed quality improves, it strengthens user trust and engagement. If it fails, it risks becoming a less credible venue for professional life, and that is a reputational and economic cost platforms feel everywhere from advertiser demand to user retention.