Everyone's using AI for content. That's exactly why it stopped working.

The platforms stopped rewarding AI content months ago. Most brands are still posting like nothing changed, and wondering why the numbers keep dropping.

For about two years, using AI to write social posts felt like a free unlock. Type a prompt, get a caption, post it, move on to the next client. Everyone in this industry did it, myself included, because it made an unreasonable workload reasonable again. The problem is that "everyone did it" is exactly the sentence that should have been a warning sign. When an entire industry adopts the same shortcut at the same time, the shortcut stops being an edge. It becomes the baseline, and then the platforms notice the baseline shifted.

That's basically what happened. Somewhere in the last year, feeds quietly filled up with content that all sounded the same: same rhythm, same "it's not X, it's Y" sentence structure, same hollow confidence. Audiences started tuning it out before they even scrolled past it. And the platforms, which exist to keep people scrolling, responded the only way platforms know how. They changed what gets distributed.

The crackdown is real, and it's more specific than people think

LinkedIn is the clearest case. Earlier this year it started actively suppressing posts that read as generic, templated AI output, not removing them, just quietly limiting how far they travel. It added a report option straight into the three-dot menu on every post, feeding a detection system it claims is already highly accurate at spotting the pattern. Instagram did something similar from a different angle, making accounts that repost or recycle unoriginal content ineligible for recommendation, a rule that used to apply mostly to video and now covers photos and carousels too. TikTok has flagged over a billion AI-generated videos through its own detection layer.

None of this is a ban on AI. That's the detail people keep getting wrong when they hear about it secondhand. What's actually being targeted is sameness. A platform can't tell whether a tool touched your draft, but it can absolutely tell whether a piece of content looks like every other piece of content in its category, and that pattern is what's losing reach now.

The real damage happens after someone reads it, not before

Here's the part that should worry brands more than any algorithm change. Roughly half of people say they engage less with content the moment they suspect it's AI-written, often before anything even confirms it. Among younger audiences specifically, about half say they've unfollowed, muted, or blocked an account for feeling like AI slop. That's not a platform penalty. That's a person deciding, in about two seconds, that an account isn't worth their attention anymore.

And it compounds. Content that reads as generic pulls in shallow engagement, a stray emoji, a one-word comment, the kind of reaction that itself reads as automated. Platforms are increasingly sophisticated enough to notice that pattern too, not just at the post level but at the comment level. Low-effort content attracts low-effort engagement, low-effort engagement signals the content wasn't worth much, and reach drops further. Nobody flips a single switch to punish you. The whole loop quietly works against you at once.

Most brands are still optimizing for the wrong problem

The instinct that got everyone into this mess made sense at the time. Posting consistently used to be the hard part, so anything that made posting easier felt like progress. But volume was never really the bottleneck, not for very long. The bottleneck now is trust, and trust doesn't scale the same way output does. You can 10x how much content a brand publishes in a week. You cannot 10x how much a stranger trusts that brand, not with the same lever.

So a lot of businesses are currently doing more work to get worse results, and blaming the algorithm instead of the strategy. The algorithm didn't turn against them specifically. It turned against the category of content they've been producing, and they haven't updated the playbook to notice.

What I actually do differently now

My rule is simple, and it's less about avoiding AI than about where it's allowed to sit in the process. AI gets the research, the structure, the first pass at organizing an idea. It does not get the actual sentences that go out under a client's name. The moment a tool writes the final draft, the voice flattens into something that could belong to anyone, and "could belong to anyone" is precisely the thing both audiences and algorithms have started penalizing.

In practice that means using AI to pull together what's already been said on a topic, to sketch an outline, to surface an angle worth arguing. Then writing the actual post myself, in a voice specific enough that swapping the brand name at the top would make it read wrong. A comment thread turned into a caption. A real client conversation reframed as an insight. A number from something I actually worked on, not a stock statistic. None of that is slower than it sounds once it's a habit, and none of it is fakeable by a tool trying to sound human on command.

Why this actually favors smaller brands

This is the part I think gets missed in most of the coverage. A large company running content through an AI pipeline at scale has a structural weakness right now: sameness is baked into how they operate. A small business or an independent marketer has the opposite advantage. A specific point of view costs nothing extra to produce. It just requires someone who's actually close to the work being willing to say something a competitor couldn't say the exact same way.

That's a real shift in leverage, not a talking point. For years, bigger budgets meant more content, and more content mostly won. Right now, more content without a real voice behind it is actively losing to less content with one. That's the opening, and it's open specifically because most brands haven't noticed it yet.

Frequently asked questions