AI slop is hitting feeds faster than anyone can scroll past it, and the pitch from the billionaires funding this generative future — Zuckerberg, Musk, Altman, etc. — hasn’t convinced their own users. It seems that people increasingly want less machine-made slop, not more of it. Now, some of the biggest social platforms are quietly offering ways to turn down the synthetic noise that they spent the last two years amplifying.

People aren’t shy about saying why they’re tired of the synthetic stuff. Fandoms are banning AI “fan art” because it flattens every character into the same glassy-eyed template, and book communities have pushed publishers into replacing covers that were quietly built from Midjourney scraps. Multiple studies keep pointing in the same direction: People want AI to help artists, not create the art; people show a clear bias against AI-generated art when they think a machine made it; and people choose human-made work even when they can’t immediately tell the difference. 

Platforms such as webtoon publishing website Tapas have already gone as far as banning AI artwork entirely, and DC Comics is reassuring fans it won’t use generative tools — “not now, not ever.” Even on Facebook Reels and TikTok, the cutest and tastiest corners of the internet feel off, because no one asked for a wave of AI puppies with six legs or recipe videos that look like they were assembled by someone who has never set foot in a kitchen. The appetite is still for the real thing.

Billions of AI videos (1.3 billion, to be exact) now sit inside TikTok’s own labeled archives, and image generators have pumped out billions more images across the broader web, enough that in communities where uploads can be tracked — like Pixiv, which said, using 2023 data, that roughly 15-20% of recent posts are AI — the synthetic stuff is clearly reshaping the feed.

So some platforms are starting to inch toward something that resembles a kill switch. 

TikTok is testing an AI-content on-off slider that quietly admits that, for some, its For You page has tipped too far into the uncanny. Pinterest just rolled out an anti-AI-slop control system that is designed for the people who want to browse handmade ceramics without wading through AI-generated living rooms that look like they were staged by an algorithm with a eucalyptus obsession. YouTube keeps asking creators to label their synthetic clips, even though those tags rarely survive a reupload, and Instagram is experimenting with disclosures after users complained that its recommendations were even more inauthentic than usual.

The slop is now hurting the product. Feeds thick with synthetic filler make people scroll less, trust less, and post less, and that equation finally matters more than the novelty of letting the machines run wild. And for the first time in a while, the platforms seem willing to admit that people might want a little less of the future they were promised.

A For You page escape route

TikTok’s AI escape hatch is a simple on-off slider tucked inside its Manage Topics menu: Settings → Content Preferences → Manage Topics → AI-generated Content. Push the AI-generated content slider down, and the app starts stripping out the synthetic clips that have been overrunning the feed — the “cinematic moments,” the AI drama recaps with faces that never quite blink right, the stitched voiceovers that sound like a podcaster trapped in a tin can. 

With the switch toggled to “off,” a feed full of BTS fancams, lip-liner reviews, and chrome-heavy home-design inspo stays closer to that lane instead of drifting into uncanny roommates, generated “storytime” romances, or AI kittens with fur that looks like wet carpet. TikTok told regulators it has already labeled more than 1.3 billion AI-generated videos on the platform, which means that if you don’t touch that control, the algorithm has plenty of machine-made filler to lean on before it shows you another human. 

TikTok is also layering in detection to try to keep that slider from turning into theater. Any video made with tools such as AI Editor Pro, or uploaded with C2PA’s Content Credentials, will carry an invisible watermark that TikTok can read, so the app can still spot AI clips after they’ve been edited, reposted, or hauled onto the app from somewhere else. That gives the company a better map of how much generative content is circulating in the For You feed at any given moment — and slider offers users at least one place to push back when their carefully tuned algorithmic mix goes awry. More Jungkook, fewer Franken-faces.

Saving your inspo boards from the uncanny

Pinterest has become ground zero for AI-warped inspiration: Hairstylists, florists, and wedding planners told the Washington Post earlier this year that AI hair, makeup, and event “ideas” were distorting clients’ expectations so badly they had to start warning people upfront. Pinterest’s pitch is that you can save something and make it real; AI broke that promise. Users have been begging for a filter for months; Pinterest responded by adding clearer “AI-modified” labels and, now, the option to blunt the feed’s synthetic edge. Reddit threads cheered when the toggle finally appeared.

The company tucked the fix under: Settings → Refine Your Recommendations → GenAI Interests, where you’ll find a set of sliders that let you push AI-generated images out of categories such as beauty, architecture, children’s fashion, art, sport, and home décor. Turn those sliders off, and the feed stops steering you toward the “dream pantry” layouts that feature IKEA hacks that defy physics. 

Pinterest’s leadership keeps saying the feature is a “see less” tool, not a “see none” guarantee, and CEO Bill Ready has argued that no platform can fully remove AI-generated content once it’s loose in the ecosystem. The website says that, “at Pinterest, we’ve always used both forms of classical AI to help people find inspiration to create a life they love.” That may be true, but the slider at least gives people a way to protect the parts of Pinterest that made the site worth opening: the handmade ceramics, the real apartments, the paint colors that actually exist in stores. The site’s AI switch is a small control in a very visual corner of the internet, but it lands at exactly the moment users started to notice that too many of their saved Pins looked like renderings for houses that can’t pass a building inspection.

A patchwork of fixes

The rest of the social media industry is scrambling to contain the AI mess in its own way. YouTube now tells creators to disclose AI edits — altered faces, voices, whole scenes — but those tags vanish as soon as a video gets reuploaded, which is why familiar synthetic clips keep bouncing around the site. Instagram introduced “Made with AI” labels after users said its recommendation surfaces felt unnervingly polished. Reddit’s moderators, exhausted by generative spam, banned AI outright in dozens of communities. 

But even as some sites are finding ways to dial the slop back, others are cranking the dial further in the opposite direction. Meta has been testing Vibes, an AI-generated video feed — which Mark Zuckerberg announced with a glossy Instagram reel full of fluffy creatures and an Egyptian pharaoh taking a selfie — has already picked up about two million users this fall. (Someone replied to the demo, “Bros posting ai slop on his own app.“)

Meta has been testing Vibes, an AI-generated video feed that signals the scope of the company’s synthetic ambitions. Mark Zuckerberg announced the service with a glossy Instagram reel full of fluffy creatures and an Egyptian pharaoh taking a selfie. Already, it’s picked up about two million users. (Someone replied to the demo, “Bros posting ai slop on his own app.”) Meanwhile, OpenAI’s Sora app, a TikTok-like competitor, treats AI-generated video as a social format in its own right; users can spin up clips of themselves, their pets, their friends, or entire imagined scenes and toss them straight onto TikTok. And tools such as Luma Labs’ Dream Machine are handing anyone with a prompt box the power to produce fully synthetic video at an industrial scale.

So while TikTok and Pinterest are testing filters to keep their feeds from drifting so often into the uncanny, a parallel universe of apps is expanding the supply of generative content faster than anyone can build a way to contain it.

The platforms can pretend this runaway flood is the natural cost of innovation, but they’re the ones who spent two years tuning the internet to reward whatever could be generated fastest. These AI filters, sliders, and disclosures won’t entirely stop the generative AI surge, but they mark a first real line in the sand. Users are saying they’re tired of feeds that feel engineered to trick them, tired of models guessing at their taste, tired of scrolling through generative eye candy that fails to satisfy no matter which “creator” made it. Platforms can keep pumping out infinite synthetic clips, but their audiences still have some say in what earns a place in the scroll.

📬 Sign up for the Daily Brief