Every week another business owner tells me they've solved their content problem with AI. They've got ChatGPT drafting the LinkedIn posts, someone hits publish, and the calendar looks full. I'm not here to tell them they're wrong for trying it. I am here to say that LinkedIn just made that shortcut a lot more expensive, and most businesses haven't noticed yet.
LinkedIn has rolled out a new detection system built on what it calls 360Brew, a 150-billion-parameter AI model, specifically to identify content that "appears to be generated by AI and lacks clear perspective." When it flags a post, that post gets shown to fewer people. Not banned, not deleted, just quietly buried. The platform is also cracking down on automated comments and has added filters so users can restrict what they see to verified profiles only.
The scale of the impact is worth sitting with. Organic reach on LinkedIn is reportedly down around 50% year over year, and 95% of users say their reach has stagnated or declined. If your business has been posting the same volume of content and getting less back for it, this is very likely why. The platform got flooded with what the industry is now openly calling "AI slop," content that reads smoothly, ticks every formatting box, and says nothing anyone would remember. LinkedIn built a model specifically to find it and starve it of reach.
If you've spent the last year having AI write your LinkedIn content for volume, ask yourself what you actually built. A content calendar, or a following?
Polish was never the point
Here's the part that should change how you think about this, not just react to it. The same reporting that documents the reach collapse also shows the opposite happening for a specific type of post: content with a genuine, specific point of view from someone who clearly knows the topic is outperforming previous benchmarks, not just holding steady. LinkedIn isn't punishing AI as a tool. It's punishing the absence of a perspective, which AI happens to be very good at producing at scale when nobody bothers to add one.
This should feel familiar, because it's the same pattern search has gone through. Google spent the last two years rewarding content that answered a real question with a specific number, a direct quote, a named source. Generic, keyword-stuffed pages got buried the same way generic LinkedIn posts are getting buried now. Every major platform is converging on the same rule: AI can help you say something faster, but it can't decide what's worth saying. That part still has to come from a person who actually runs the business, closes the deals, or fixes the problem.
This isn't a LinkedIn quirk. It's the new rule
If LinkedIn were the only platform doing this, you could file it away as one algorithm change to work around. It isn't. TikTok has just updated its live shopping rules to ban non-real-time verbal interactions, including AI-generated voices, meaning hosts running a livestream sale now have to actually be there, talking, in real time. New York has introduced a law requiring businesses to disclose when an advertisement uses an AI-generated performer, and industry watchers expect other states and eventually national rules to follow the same path. Three different platforms and regulators, in the same month, arriving at the same conclusion: the audience has to know, or be able to tell, when they're dealing with a real person.
None of this is really about AI being bad. It's about what happens when a genuinely useful tool gets used as a replacement for judgement instead of a support for it. Every platform that built its business on people wanting to hear from other people is now recalibrating for a flood of content that technically works but has nobody actually behind it. The businesses that get caught out aren't the ones using AI. They're the ones who let AI make the decision about what to say, instead of just helping them say it.
What this means for your content this week
You don't need to abandon AI tools. You need to stop using them to skip the one step that actually matters.
- Before anything goes out, it should contain an opinion someone could disagree with. Not a summary of a trend, a position on it.
- It should include something only your business would know: a real number from a real job, a specific client situation (with permission), a decision you made and why.
- If you can swap your company's name for a competitor's and the post still reads the same, it's the kind of content LinkedIn is now built to bury.
AI scales what's already there. If what's there is a genuine point of view, AI will help you publish it faster and more consistently than you could alone, and right now that's rewarded more than ever. If what's there is nothing, AI just helps you publish nothing faster, and now the platform is built to notice. The business owners who treat this as a five-minute content briefing exercise once a week, not a prompt they forward to an assistant, are the ones whose reach will keep climbing while everyone else wonders where their audience went.