Every major platform now has an AI slop policy. Read them carefully and a pattern shows up: almost none of them reduce how many people see AI-generated content. They either tell you it's AI, or hand you a dial to turn it down yourself.
Those aren't the same thing, and the gap between them explains why your feed hasn't noticeably improved.
Four different responses, one label
Disclosure is the most common. Meta puts an "AI info" label on content its systems flag as AI-generated, and says outright that labeled content isn't penalized in ranking for carrying the label. TikTok makes creators label AI content and adds an invisible watermark to the metadata of some videos. Pinterest lets creators mark work as AI-modified or generated and runs detection of its own.
User controls come second. TikTok added a toggle for dialing down how much AI content shows up in your feed. Pinterest built something similar. The platform isn't deciding what deserves distribution — it's giving you a knob and backing away.
Monetization limits are third. YouTube has cut monetization for AI-generated rip-offs of other creators' work and says it'll expand detection and labeling of low-quality AI video, all while shipping more generative tools of its own.
Actual reach suppression is fourth, and it's rare. Snapchat stopped recommending fully AI-made videos in Spotlight, which cuts an amplification path rather than the content itself. LinkedIn is the only major platform with a confirmed policy that demotes reach based on how a post reads: since May it has suppressed — not removed — formulaic AI posts and engagement bait, holding flagged content largely to the poster's first-degree connections. It's also testing a way to privately flag posts and comments as AI.
Most coverage has lumped all four together as "platforms are downranking AI." Only the last one is.
Why the soft options keep winning
The conflict here isn't subtle. The companies being asked to suppress synthetic content are the same ones handing out the tools that make it, and generative features are among the products they most want people using. A policy that suppressed AI content by origin would suppress output from features the platform is busy promoting.
There's a real measurement problem too. Detection isn't reliable at scale. False positives punish people who used AI to fix their grammar or generate a background, and nobody wants to explain to advertisers why legitimate posts disappeared.
LinkedIn dodges part of this by targeting a quality signal — formulaic writing, engagement bait — instead of AI origin. That's easier to defend, and it catches human-written slop as a bonus.
If you post things
The practical rules are simple. Label AI use where it's required, because undisclosed use is what actually triggers enforcement. Assume metadata watermarks travel with your file wherever it goes. And expect generic, templated writing to underperform on LinkedIn no matter who or what produced it.
If you mostly read
Use the controls that exist. The feed toggles on TikTok and Pinterest are the only tools currently on offer that meaningfully change what lands in front of you, and most people have never opened those settings once.
The bigger question is still hanging. Labels tell you what you're looking at. They don't reduce how much of it there is. Until a platform is willing to distribute less of something it also profits from, disclosure is where this crackdown tops out.
Image: Towfiqu barbhuiya, via Pexels
