Key Takeaways
- Kapwing found 104 of the first 500 videos recommended to a brand-new YouTube account were AI slop, and 278 channels built entirely on it have logged 63 billion views and roughly $117 million a year in ad revenue.
- Animoto's State of Video 2026 report finds 83% of consumers say they have watched a video they suspected was AI-generated, and 36% say a suspected-AI brand video lowers their opinion of the brand behind it.
- 95% of marketers say their own branding must be visible in AI-assisted video and 90% insist on the ability to edit it themselves, according to the same report.
- IAS and YouGov found 83% of US digital media experts expect brand safety concerns to keep rising, yet 61% say they are excited to advertise inside AI-generated content and only 2% reject the adjacency outright.
Two things are true about AI video at once, and neither is going away. Marketing teams are pouring budget into generative tools that make short-form video cheaper and faster to produce than it has ever been. Audiences, meanwhile, are getting better at spotting the result, and a growing share of them do not like what they see. The gap between those two facts has just acquired a name, a body of data, and a line item on the media plan: brand safety.
How Much of the Feed Is Already Synthetic
The clearest measurement comes from a report from the video-editing company Kapwing, which catalogued the top 100 trending YouTube channels in each of several countries and then opened a brand-new account to see what the platform's own algorithm would recommend it. Of the first 500 videos pushed into that empty account's feed in October 2025, 104, just over one in five, met the researchers' definition of AI slop: low-quality, auto-generated content built purely to farm views rather than to inform or entertain anyone.
The scale behind that single feed is not a rounding error. Out of 15,000 of the world's most popular channels, 278 were found to run exclusively on AI-generated content. Between them those channels have logged more than 63 billion views and 221 million subscribers, and Kapwing estimates they collect roughly $117 million a year in ad revenue. This is not a fringe experiment running out of view of advertisers. It is a functioning, monetised segment of the same platforms brands already buy media on.
What Viewers Actually Notice
The volume of synthetic video would matter less if audiences could not tell the difference. They increasingly can. Animoto's State of Video 2026 report found that 83% of consumers say they have watched a video they suspected was AI-generated, flagging robotic gestures, unnatural voices, and a flat emotional tone as the giveaways. More consequential for marketers, 36% say a suspected-AI brand video actively lowers their opinion of the brand behind it.
Marketers are not blind to this. Animoto's same survey found 90% say it is essential that they can edit AI-generated content themselves, 95% insist their own branding has to be reflected in it, and 99% say the brand's personality has to come through. "With the rise of AI, marketers can supercharge how they create and scale video," Animoto CEO Beth Forester said of the findings, "but they also risk eroding the authenticity that builds trust." Separate research cited by eMarketer found 85% of consumers notice uncanny-valley tells in synthetic content, and nearly half would cut back on a social platform altogether if AI content kept increasing in their feed.
A Brand Safety Problem Nobody Has Solved Yet
Ad buyers are not ignoring the shift, but they are not aligned on a response either. Research from IAS and YouGov, reported by eMarketer, found 83% of US digital media experts expect brand safety concerns to keep climbing as video ad volume grows, and 53% already name proximity between their ads and generative-AI content as a top media challenge for 2026. Yet the same research found 61% of those experts say they are excited to advertise inside AI-generated content, with only 2% rejecting the adjacency outright, a split that looks less like a policy and more like an industry still arguing with itself.
Digiday's reporting on the topic captures the same uncertainty from inside agencies. "This is going to end up being the next big brand safety problem," Zefr chief commercial officer Andrew Serby told the outlet, while most brands remain in what the piece calls watch-and-wait mode, drafting internal frameworks rather than shipping firm rules. DEPT's editorial director Salazar Llewellyn put the working principle more simply: "Our approach is always human-led, craft-first, augmented with AI. You can't automate good judgment." The same reporting found marketers rate user-generated content as extremely important to their social strategy 18 times more often than they rate AI content that way.
Put together, the numbers describe a market moving in two directions simultaneously. Production has never been cheaper or faster, which is exactly why the volume of low-effort synthetic video keeps climbing. Trust in that same video is moving the opposite way, and it is moving fastest among the viewers brands most want to keep. The teams treating this as a governance question, not a production shortcut, are the ones positioned to come out ahead.
- Write down what "AI-assisted" is allowed to mean. Animoto's data suggests audiences forgive AI voiceovers and background editing but punish fully synthetic content, so the line between the two needs to be a documented standard, not a judgment call made per project.
- Keep a human in the edit, not just in review. With 90% of marketers already insisting on edit access to AI output, formalise that as a workflow requirement rather than an assumption someone downstream will catch a problem.
- Audit ad adjacency on video platforms. With more than one in five recommended videos on some platforms now AI slop, confirm where placements actually land rather than trusting brand-safety settings you have not re-checked this year.
- Protect your library of real human footage. As synthetic video gets cheaper, authentic footage becomes the harder-to-fake asset, worth budgeting for deliberately rather than treating as something to cut first.
- Decide where automation actually saves time. Not every production step benefits equally from AI tooling, and teams that map the workflow honestly protect the moments that need a person and speed up the ones that don't.
- Brief creators and agencies on your standard, in writing. A private internal policy does not travel to every freelancer or partner touching your video output; an explicit brief does.
