Social platforms spent the past several years making it easier to create content with AI. Now, they are trying to limit the low-quality, repetitive content those tools can produce.
Since March, Facebook, TikTok, YouTube, LinkedIn, Snapchat and X have announced or clarified policies that favor original work, restrict mass-produced content or change which creators can earn money. None of these platforms is banning AI outright. Most still allow creators to use it for editing, animation, research and other production work.
The policies focus on a different problem: content made at scale with little original thought, information or creative input.
Snapchat removes fully AI-generated videos from recommendations
In July, Snapchat said wholly AI-generated videos would no longer be eligible for recommendations on Spotlight, its public short-form video feed.
Spotlight recommendations allow creators to reach people who don’t already follow them. Losing that eligibility limits how widely a video can be distributed through Snapchat. The company said it wants Spotlight to prioritize videos made by people, including personal stories and original perspectives.
Snapchat will continue recommending videos that creators edit or enhance with its AI tools. Those videos will include labels explaining that AI was used. Snapchat allows AI as part of the production process, but it doesn’t want fully generated videos filling its recommendation feed.
X shifts creator payments toward original work
While Snapchat is changing what gets recommended, X is changing what gets rewarded. On September 8, the platform will replace its revenue-sharing program with Original Content Rewards.
The program will pay creators for qualifying views on original reporting, analysis, photos, videos, illustrations and other work they produce.
Creators can still comment on someone else’s work, but they must add meaningful information or analysis. Downloading and reposting another person’s video won’t qualify. Neither will minor changes designed to make copied content look original.
X hasn’t promised that every original creator will earn more money. The change affects which content qualifies for payment, though individual creators aren’t guaranteed to earn more.
It also hasn’t explained how it will decide whether AI-generated content contains enough original work to qualify. X has separately said that an updated version of Grok can detect duplicated videos and text more effectively, including copies with added watermarks or minor edits.
The system may be effective at identifying stolen content, but determining how much human work went into an AI-generated video is a different challenge.
X hasn’t released details about how it will make that judgment at scale or how creators will appeal incorrect decisions. For now, the program shows X placing greater value on original work while still allowing AI-generated content.
Facebook reduces the reach of unoriginal videos
Meta is applying the same principle to distribution. In March, the company said it would give original creators more reach and monetization opportunities across Facebook Feed and Reels, while reducing the reach of unoriginal content.
Facebook considers a video original when the creator filmed or produced it. Creators can also use material from other sources if they add new information, analysis or substantial changes to the story.
Small edits don’t qualify. Adding captions, changing the playback speed or placing a border around someone else’s video isn’t enough.
Accounts that repeatedly publish unoriginal material can lose access to recommendations and monetization. The policy applies to copied content whether or not AI was used. However, it could affect automated workflows that combine existing clips with AI narration, captions or other minor changes.
YouTube clarifies its monetization rules
YouTube has restricted repetitive content for years. In July, the company clarified how its inauthentic-content policy applies.
The update gave creators more detail about the content that can make a channel ineligible for the YouTube Partner Program, which allows creators to earn money from advertising, subscriptions and other features.
YouTube can remove monetization from channels filled with generic, repetitive or template-based videos that show little variation.
The guidance also covers content designed to provoke emotional reactions mainly to generate views. It places additional restrictions on AI personas presenting information about sensitive subjects such as health and finance.
Using AI doesn’t automatically make a video ineligible for monetization. YouTube is focused on channels that publish large numbers of similar videos with limited creative or informational value.
TikTok tests AI-spam detection
TikTok is taking a more limited approach. In July, the company said it would test improved detection systems for accounts that post AI-generated spam.
The initial test covers politics, current events, financial advice and medical content. TikTok selected those subjects because inaccurate information could affect public trust or well-being.
The test doesn't apply to every AI-generated video on TikTok. It targets accounts using AI to mass-produce spam in several sensitive categories.
TikTok also uses creator disclosures, digital Content Credentials and watermarking technology to identify AI-generated videos. The company says it has labeled more than 3 billion videos using those systems.
The company continues to provide AI creation tools, even as it builds systems to identify how those tools are being used. For now, its response is targeted rather than platform-wide, focusing enforcement where mass-produced AI content carries greater risk.
LinkedIn asks users to report AI slop
LinkedIn is addressing a similar issue with written posts. The company added a “Seems like AI slop” reporting option for posts and comments. LinkedIn will use those reports to improve how it identifies and reduces the distribution of low-quality content.
It’s also replacing its AI-powered post-rewriting feature with a proofreading tool that makes smaller corrections without changing the writer’s voice.
LinkedIn is less focused on video than the other platforms, but its changes point to the same concern.
Making content easier to produce can also make feeds more repetitive and less useful.
Audience data from RAD Intel shows how this dynamic is playing out among LinkedIn marketing audiences. One audience, Martech Innovators, actively explores generative AI and automation tools to improve their workflows, while another, Authority Builders, invests in publishing original industry perspectives that build credibility with decision-makers. The findings show that interest in AI efficiency can exist alongside continued investment in original expertise and perspective.
What this means for brands
For brands, these changes put more pressure on how AI is used throughout the creative process. AI can still support research, editing, translation, visual effects and other production work while leaving the core idea, perspective and creative judgment with people.
The greater risk comes from automating the entire process. If AI generates the concept, script, visuals and narration with little input or review from a person, the result is more likely to resemble the repetitive content these platforms are targeting.
Brands should be able to identify what they contributed to each piece of content. That contribution could be original footage, firsthand expertise, a customer story, a product demonstration, new information or a clear opinion from someone inside the company.
There’s also a larger implication for how brands decide what to create. AI has made content production faster and cheaper, which means brands can produce far more than they could a few years ago. But as platforms become more selective about what they distribute and reward, producing more content becomes less valuable if teams don't understand what their audiences actually want to see.
Audience understanding, creative judgment and original perspective matter earlier in the process. Brands need a reason for creating something before AI helps them create it faster.
