LinkedIn Fights Back Against AI Slop With New Detection Tools

LinkedIn is drawing a line in the sand against what’s become an epidemic: AI slop. The company announced Thursday that it’s rolling out a “seems like AI slop” button, allowing users to flag posts that appear to have been churned out by AI rather than written by actual humans. It’s a small gesture that signals something bigger: platforms are finally fed up with the flood of low-quality, computer-generated drivel clogging their feeds.

The problem has become impossible to ignore. Bot traffic now exceeds human traffic on the web, according to Cloudflare, a milestone the infrastructure firm reached faster than expected. Reddit competitor Digg had to shut down earlier this year because it couldn’t manage the bot infestation. Even Substack, the newsletter darling, just partnered with an AI detection company called Pangram to help users identify AI-written content.

The Real Problem Behind the Slop

LinkedIn’s Chief Product Officer Hari Srinivasan didn’t mince words about the stakes: “AI slop is a top priority for all of us. We really care about this. People come to LinkedIn to connect with real people and share their real perspectives, ideas, and expertise.”

That’s the core issue right there. Social media platforms built their value on human connection and authentic voices. When your feed gets swamped with AI-generated motivational platitudes or faux-expert takes, the whole proposition falls apart. LinkedIn users aren’t there to read what a language model thinks a productivity guru would say. They’re there to learn from actual professionals.

The new button is just one piece of LinkedIn’s multi-pronged approach. The company is already blocking hundreds of thousands of automated comment attempts daily and millions more over just a couple months. Behind the scenes, LinkedIn is deploying new classifiers trained to identify both AI slop and other low-quality content, which will reduce what you see in recommended posts from outside your network.

A Carrot and a Stick Approach

Here’s where it gets interesting: LinkedIn is distinguishing between people using AI tools to enhance their own writing versus those posting pure slop. For the former group, the company will quietly flag inauthentic-seeming content in user dashboards. The idea is to nudge people toward better writing practices without publicly shaming them.

It’s also scrapping its own “enhance your post” AI feature that used to rewrite your words entirely. That’s being replaced with a proofreading tool that only catches typos and grammar issues, preserving your voice instead of replacing it. That’s a meaningful shift. LinkedIn acknowledged that its own AI writing assistant was part of the problem.

Other improvements include expanding verification tools and letting users block comments from company pages they don’t want to see. These are smaller fixes, but they all add up to a platform trying to reclaim some semblance of authenticity.

The Bigger Picture

LinkedIn’s moves reflect a broader awakening across tech platforms. People are exhausted by AI slop. They’re tired of feeds filled with generic, soulless content that feels like it was written by an algorithm (because it was). The backlash is real, and platforms are starting to listen.

Pangram just raised $9 million specifically to combat AI content flooding the internet. That’s a vote of confidence that this is a solvable problem, or at least a manageable one. But solutions require platforms to actually care, and they require users to call out the garbage when they see it.

LinkedIn’s button gives users a voice in the fight. It’s crowdsourcing moderation in a way that might actually work. Every flag helps train LinkedIn’s AI models to get better at spotting slop without human review. It’s a virtuous cycle if it works.

The real test will be whether these tools actually improve the user experience or just create more noise. If LinkedIn can meaningfully reduce low-quality AI content while preserving authentic AI-aided writing, it might have found the balance. But that’s a delicate line to walk.

Source: Infeeds.com

Will platforms ever fully solve the AI slop problem, or are we just buying time before the bots find new ways to game the system?