Snapchat Says That AI-Generated Content Will No Longer Be Recommended In Spotlight

Snap Inc. announced that wholly AI-generated videos will no longer be eligible for recommendation on Spotlight, the short-video feed in Snapchat, effective this month, to reward authentic human-made content. The company reported a 120% increase in unique Spotlight contributors globally compared to last year. Snap's move follows similar actions by LinkedIn, YouTube, and Spotify to curb AI-generated content, with LinkedIn adding a 'Seems like AI slop' reporting option and YouTube clarifying its inauthentic content policy.

There seems to be a bit of a brewing backlash against AI-generated content by major platforms. Snap has announced that wholly AI-generated videos will no longer be eligible for recommendation on Spotlight, the Discover-adjacent feed where Snapchat surfaces short vertical videos to a broad audience beyond a user’s friend list. The company framed the move as an effort to reward “authentic, human-made content and help original creators reach an audience,” and said the change is already in effect this month. In a post titled “Rewarding Authentic Creativity on Spotlight,” Snap said its recommendation systems will increasingly favour original, human-made creativity over AI output, building on an ambition it laid out earlier this year to keep Spotlight feeling real. The company pointed to a surge in participation as evidence the strategy is paying off, saying the number of unique Spotlight contributors globally is up more than 120 percent compared to last year. Snap was careful to distinguish this from a blanket rejection of AI tools. Content that has been edited or enhanced using Snapchat’s own AI creative features will still be eligible for recommendation, and will carry transparency indicators so viewers know AI played a role. What gets excluded is content that is wholly generated by AI, the kind of video with no human hand in the actual creation, as opposed to a Snap filter or AI-assisted edit layered onto something a person shot themselves. The company also acknowledged its own limits here, noting that no detection system is perfect. That’s a fairly candid admission for a platform announcement, and it points to the same underlying problem every company wrestling with this issue keeps running into: telling wholly AI-generated content apart from AI-assisted content at scale isn’t a solved problem, and probably won’t be for a while. Snap joins a growing list Snap’s announcement lands the same week LinkedIn rolled out its own response https://officechai.com/ai/linkedin-now-lets-users-report-posts-as-seems-like-ai-slop/ to the same problem, though its approach looks quite different. LinkedIn added a “Seems like AI slop” reporting option to its post and comment menus, alongside classifiers meant to catch low-quality AI content before it spreads, and a private dashboard notification telling users when their own posts came across as inauthentic. LinkedIn’s Chief Product Officer Hari Srinivasan said the platform is catching hundreds of thousands of automated comment attempts a day. Notably, LinkedIn also killed its “enhance your post” AI rewriting feature and replaced it with a plainer proofreading tool, essentially conceding that its own AI writing assistant had been contributing to the sameness problem it’s now trying to fix. YouTube has been on a similar track for a while, though it has approached things from the monetization side rather than recommendation. In July, the platform clarified its “inauthentic content” policy, first introduced last year, into three specific categories that can’t earn ad revenue: generic or template-based videos churned out at scale, emotionally manipulative content, and AI personas discussing sensitive subjects like finance or health. YouTube has repeatedly said this isn’t a ban on AI-assisted content, just on channels built around mass-produced, low-effort output. A creator using AI as part of a genuine creative process, with a real point of view attached, is still fine. A channel running templated scripts through text-to-speech at volume is not. Spotify has faced a version of this fight too, though its stance has been messier. The company said it removed more than 75 million “spammy” AI-generated tracks from its catalog over the past year, while still stopping short of banning AI music outright or labelling it consistently, which has left third-party trackers like SlopTracker stepping in to flag suspected AI artists that Spotify itself won’t identify. Deezer and Tidal have gone further, moving to demonetize AI-generated music rather than simply policing fraud around it. TikTok, meanwhile, has leaned into automatic labelling using C2PA content credentials, and says it has tagged well over a billion AI-generated videos, though its own detection still misses a majority of undisclosed AI content by some estimates. What ties all of this together is that none of these platforms are trying to ban AI outright. Every one of them has drawn the same line: AI as a tool in a human’s creative process is fine, AI as the entire process is not. The harder question, one none of them has really answered yet, is how reliably any of this can actually be enforced once the volume of AI-generated content outpaces what any detection system can keep up with. For Spotlight specifically, the practical effect is that creators leaning entirely on AI generation tools for their videos will need to either add a real human layer to what they make or accept that their content simply won’t be pushed to new audiences. Existing content already on the platform doesn’t appear to be at risk of removal, only future recommendation eligibility.