Tech Leaders Urged to Curb AI-Generated Content

It’s becoming nearly impossible to scroll through any major social platform without encountering AI-generated content, but that doesn’t mean users have to accept it as inevitable. Over the past year, platforms like YouTube, Instagram, TikTok, and others have stepped up their content authentication efforts, rolling out automatic labels meant to distinguish synthetic images, videos, and music from work produced by actual humans.

That’s a fine step if you happen to stumble across a labeled post at random. But a far more practical solution would be letting users filter out AI-generated content entirely. Right now, even with labels in place, the experience of browsing online hasn’t changed in any meaningful way. You might notice a TikTok or YouTube video now includes an AI disclosure in the description or a small information badge overlaid on the clip. Meta does something similar, slapping “AI info” labels on Facebook and Instagram images that carry identifying metadata or voluntary creator disclosures.

But if you want to actively avoid seeing anything tagged with those labels , a reasonable desire given the cognitive drain, ethical concerns, and environmental toll tied to generative AI , you’re out of luck. It’s shockingly difficult. A simple toggle, an “AI” checkbox, would solve the problem instantly.

I reached out to Meta, Google, TikTok, and Spotify to ask whether they have any plans to let users filter content they’ve already been authenticating with AI labels. TikTok and Spotify never responded. Google said it had nothing to share. Meta declined to provide an attributable comment. In short, none of them said yes.

One of the only platforms I’ve found with an AI content filter is DeviantArt, and its implementation speaks volumes. You can’t access it from feeds or the store page. Instead, you have to create an account, hover over your user icon in the top-right corner, and dig into the “AI Content Settings” menu. From there, you get exactly two options: the default “Show AI” setting, or “Suppress AI,” which claims you’ll see “fewer instances” of AI-generated or manipulated imagery.

I tested both settings. Honestly, I couldn’t tell much difference. I’ve gotten pretty good at spotting AI-generated “digital illustrations,” but I didn’t have to rely on my eye alone , nearly every suspicious image I clicked had a creator disclosure confirming it was machine-made. DeviantArt does a poor job of automatically applying AI labels to images whose metadata clearly indicates AI origins.

Pinterest has a similar system. Signed-in users can click the settings icon, select “Refine your recommendations,” then tap the “AI content” tab to toggle categories like art, beauty, fashion, and home decor. According to Pinterest, disabling any of these will show you “less AI-modified content” for that category. In my experience, it’s far from effective. The setting is also buried deeper than a feed-based filter should be. Even with AI filters maxed out, I still saw plenty of images with suspicious tells , uncannily perfect models and unexplainable illustration errors.

I appreciate the customization, but these refinement options are hidden and don’t actually work well. And that’s almost certainly what would happen if YouTube or Instagram introduced an AI content filter: it wouldn’t work very well. But that’s okay, because it would expose the ineffective “solutions” our AI emperors parade around in. These features exist on paper to appease regulators and critics, but they do little to solve the real problem of distinguishing AI fakery from authentic photography and creative work.

Platforms know it’s a problem. Instagram head Adam Mosseri said in December that “authenticity is becoming a scarce resource” as AI-generated content rises. Google CEO Sundar Pichai admitted in a recent Decoder interview that “there’s a lot of AI slop out there,” and that users need to “adapt to it.” Fine , give us filters.

Provenance-based systems like C2PA and SynthID work by embedding metadata or invisible watermarks at the point of creation. But plenty of open-source AI models don’t use these (especially those designed for malicious purposes), and metadata can be stripped too easily to be reliable. Detection-based methods analyze patterns in digital content and rate the likelihood of AI involvement, but they can generate false positives. None of this works effectively at scale right now.

Still, companies like OpenAI are touting these labeling solutions as a way to prevent people from being duped by deepfakes and other misleading content. If regulators fully understood how ineffective they are, online platforms and AI providers might be forced to find a real solution instead of what feels like a smokescreen.

Platforms will argue they risk incorrectly flagging authentic content if they push labeling too aggressively. Both Meta and YouTube learned this the hard way after applying AI labels to content creators insisted was made without such tools. If that’s a genuine concern, then find a better solution. Surely improving the user experience for millions of users is a worthwhile investment to stay competitive.

And while I’m at it, why can’t I report all the unlabeled AI slop I see every day? Given the scale , a Kapwing study last year found that more than 20 percent of YouTube videos shown to new users is low-quality generated slop , a lot of human moderators would be needed to vet each report.

And maybe that’s the real issue. At a time when big tech is replacing workers with AI that supposedly outperforms them, can they afford to backtrack on that narrative by hiring humans back to fix AI’s problems? Humans tend to require things like salaries and benefits, unlike automated moderation systems that lack nuanced investigative skills.

An alternative to labeling AI content would be to start labeling verified human creators instead. That wouldn’t catch synthetic content posted by those creators, but it could help us see less from unverified content farms churning out low-quality slop. That’s the future Instagram’s Mosseri pitched for Meta’s image-sharing platform, and something Spotify already does with Verified artists.

Sounds great. If only there were a way to exclusively surface music by verified artists.

Of course, Meta, Spotify, and Google don’t just host AI-generated imagery, ads, and music , they also make the tools that create it. That’s why they insist not all AI content is slop, and that it’s really a quality issue. If it gets convincing enough, they’re betting you won’t notice and will keep happily consuming. Allowing users to filter it out would undermine the massive effort these platforms have made to profit from AI. They want you to embrace the slop factory.

I’d love to be proven wrong. I’m practically begging online platforms to show that AI labeling efforts haven’t been a waste of time. But right now, they hold all the cards, and we’re left hoping their AI moderation works. So give us a basic “no AI” or “verified human creator” filter, and we’ll judge for ourselves how well this is actually going.