YouTube CEO Acknowledges 'AI Slop' Problem, Says Platform Will Curb Low-Quality AI Content (blog.youtube) 54
YouTube CEO Neal Mohan used his annual letter to creators, published Wednesday, to outline an ambitious 2026 vision that embraces AI-powered creative tools while simultaneously pledging to crack down on the low-quality AI content that has come to be known as "slop."
Mohan identified four AI-related areas that YouTube "must get right in 2026." The platform is working on tools that will let creators use AI to generate Shorts featuring their own likenesses and to experiment with music. "Just as the synthesizer, Photoshop and CGI revolutionized sound and visuals, AI will be a boon to the creatives who are ready to lean in," he wrote. Features like autodubbing, he says, will "transform the viewer experience."
But "the rise of AI has raised concerns about low-quality content, aka 'AI slop,'" he wrote. YouTube is building on its existing spam and clickbait detection systems to reduce the spread of such content. He also flagged deepfakes as a particular concern: "It's becoming harder to detect what's real and what's AI-generated." The platform plans to double down on AI labels and introduce tools that let creators protect their likenesses.
Mohan identified four AI-related areas that YouTube "must get right in 2026." The platform is working on tools that will let creators use AI to generate Shorts featuring their own likenesses and to experiment with music. "Just as the synthesizer, Photoshop and CGI revolutionized sound and visuals, AI will be a boon to the creatives who are ready to lean in," he wrote. Features like autodubbing, he says, will "transform the viewer experience."
But "the rise of AI has raised concerns about low-quality content, aka 'AI slop,'" he wrote. YouTube is building on its existing spam and clickbait detection systems to reduce the spread of such content. He also flagged deepfakes as a particular concern: "It's becoming harder to detect what's real and what's AI-generated." The platform plans to double down on AI labels and introduce tools that let creators protect their likenesses.
reality check (Score:3)
I can't remember which event it was, but someone from youtube mentioned that when generative video AI came to the scene, youtube's data ingestion went up many times higher than normal.
To the point where they started to max out their data storage and had to start culling things.
This is likely more of that.
urgh (Score:1)
It's bad enough that I have to see this crap in my feed. I used to be happy to watch a video from a creator with like 1-2 thousand subs, if it was any good I'd leave a comment to boost engagement etc. But now it's pretty much a certainty that that kind of creator is AI slop and I see so many more of those in my recommendations than ever before. It's infuriating, it's just like spam in a mailbox.
But now youtube is literally deleting videos to make room for this stuff?? Seriously?? There are so many videos wh
Re: (Score:2)
Create easier tools, get more people produce content. And people with lower skill, so don't expect great content just because they can produce content at all.
Re: (Score:2)
On the other hand, there's now a lot of content by very creative people who lacked specific technical ability needed to manifest that creative talent into something material.
It's why suno.ai is as popular as it is. It's why we now have so many fan made short films. Etc.
Re: (Score:2)
Yes. AI does not have a quality problem. Sites have a filtering problem. Creative output increased by orders of magnitude, but many sites do not manage to rank it correctly. Automatic methods are unreliable and user-rating based methods do not scale well with increased quantity of content available.
Maybe some sites should get more recommendation-based systems where users can curate lists of content they like instead of just listing content sorted by rating and view count.
Re: (Score:2)
Re: (Score:1)
Create easier tools, get more people produce content.
Why?
Re: (Score:2)
Because it is easy. There is also a step up in quality from what they could do before. Uploading a simple generic AI image has more aesthetic than uploading a paint sketch that doesn't look good yet. I'd think people are less selfconscious, when the image looks like professionally drawn, no matter if it is low-quality in terms of being generic.
Re: reality check (Score:2)
Bring back old YouTube with its non-AI slop (Score:5, Interesting)
I miss old YouTube. Back when "Broadcast Yourself" was a slogan. Back when people were making 3 to 7 minute videos to match an attention span raised on the Flash animations that preceded it, as opposed to trying to hit the magic 8:01 mark that allows midroll advertisements. Back when "slop channel" meant a channel full of a pig eating slop [youtube.com].
Re: (Score:3)
Re: (Score:2)
"transform the viewer experience." (Score:4, Insightful)
will "transform the viewer experience."
You mean.. from "I only use you on friday night on a phone call with an old friend, and too look up how to fix my car / washer / dryer whatever" to "Not using you at all?"
Seriously. Slashdot Collective says 'eat the rich?' I say "fuck all the techbros."
Yes, I'm narrowing down the carnage to just the techbros (and tech sisters, like that nutjob that headed Theranos)
They've unleashed a new cancer on society, by amplifying voices that were best left handing out flyers at street intersections. Not everyone deserves to have the biggest bullhorn in the world.
Re: (Score:1)
Meh... 'eat the rich' sounds better, though, then we could get someone who hates AI in the company, and we could get it back to what it used to be.
solution: more slop? (Score:2, Insightful)
Re: (Score:2)
Fix the ADs!!! (Score:1, Interesting)
I don't mind AI content (There's some good StarWars stuff that regular people can make now that is pretty good, tbh). But the Joe Rogan ads for crap needs to go!
If Youtube can make money by showing an ad then a real person should have to look at it first. How hard would that be? smh...
Re: (Score:2)
There is some good slop out there too. I don't know how to quantify quality slop, but there is at least one channel putting out some very funny videos that are complete nonsense. I would not want to see that go away but I don't think they are going for detecting AI vs not anyway. They had a bad slop problem before AI.
Re: (Score:2)
Yeah, well worth the cost of YouTube premium. At least for me
Re: (Score:2, Informative)
Firefox + uBlock Origin still works fine. I'm still unsure why Chrome is so popular, there never was a time when it was better than Firefox.
FWIW though, if you're getting Joe Rogan ads, it's because of your history, probably watching similar content. It is always worth, from time to time, clearing your cookies and logging in again (or starting a new account if they're tying ads to logins, which right now they don't appear to be, but are clearly preparing to do.)
Re: (Score:2)
^ This is the way. The way to fix ads was to cut ads out of our lives. It worked spectacularly well, IMHO. The only catch is that when you do it, it just works for you. So you've got to advocate it to others, if you're worried about what other people might do while under the influence of ads (i.e. what if they vote?).
Re: (Score:2)
YouTube has adverts?
You mean the incessant "and here are some words from Brilliant/ some "educational toy company"/ ... actually, I can't bring another example to mind - I FF past them so fast, without engaging the brain.
Am I destroying YT? Maybe, maybe not. Do I care? Not in the slightest.
Re: (Score:2)
Firefox and uBlock origin kind of works on slashdot. You still see the ad but uBlock prevents it from playing. Some times the ad doesn't show or play but you can't watch the video itself for a short time, 10 to 20 seconds. Often I reload the page and then I can play the video quicker. And about half the time the ad loads but won't stat playing because uBlock blocked that. And they wonder why FreeTube is so popular!
Re: (Score:2)
I mean kind of works on YouTube. Silly brain auto word suggest failure.
Re: (Score:2)
I don't mind AI content as long as it it's clearly and openly labeled as such. The Star Wars stuff is neat, albeit being often chatty, slightly off-putting and legally dubious. So much slop out there tries to pass unreal things as real and the comments that follow from gullible viewers set my hopes for humanity even lower than they already were.
Fiction and satire used to be easy to spot, but as the line continues to get fuzzier, so does critical thinking.
We need to stop AI slop. By creating more AI slop. (Score:3)
How did this guy keep his head from splitting in two the way he's talking out both sides of his mouth? Stop AI slop by creating more AI slop? What? Do these fools even listen to themselves anymore?
Re: (Score:3)
You've got to read it carefully.
It's the "low-quality AI slop" they want to crack down on. They're actively pursuing AI systems that generate video on demand, either unique to each viewer or specific to demographics.
They don't like the "low-quality AI slop", they want the high-quality AI slop, the AI-generated videos generated specifically for that individual that matches the algorithm.
And it's clear to everyone they want to be the ones generating the "high-quality AI videos" to cut the content creators
Re: (Score:3)
Exactly. They're not dumping AI content. They're doubling down on making AI content indistinguishable from content generated by humans.
Re: (Score:2)
The only way to stop a bad guy with AI slop is with a good guy with AI slop.
Low effort (Score:5, Interesting)
So, saying they are "cracking down" is kind of funny, as they aren't even doing the bare minimum enforcing their existing policies.
Re: (Score:2)
Re: (Score:1)
Like those "true crime" AI-narrated videos? "IN 1986, this young man killed his dog and parents. Blah, Blah. You won't believe what he used!" (it was a handgun... who woulda guessed it)
Once a system is in place to ID the 'slop', whatever labels it as slop will be changed in a heartbeat by the poster so that they can get around that filter.
Fake Ads (Score:2)
The majority of ads I see on YouTube these days are AI-slop deepfake celebrity endorsements, obviously-fake (and dangerous) medical scams, weight-loss salts, workout plans, etc. etc. I don't see anything in the CEO statement about that.
Give me a button that i press (Score:4, Insightful)
And it pops up a menu where i can select both subtitles and voice language, and make this button remember the setting for all videos.
Is this THAT hard?
Re: (Score:2)
I think that is there in the browser version, but you must be in the correct mode to see the button that leads you to the correct setting. Having said that, I too have had to reset settings on occasion. Once a month or so.
Re: (Score:2)
It has the options, but they're all spread all across the interface, and the auto dub thing is hidden in embedded videos unless you set em to fullscreen.
They already have the CC button that could very well be a overall language menu instead.
No incentive to stop AI slop (Score:2)
Trouble is AI slop is making YouTube a ton of money showing ads. Even if all the real creators were driven away, AI slop would still be bringing in the money. The trouble is most of YouTube's users don't seem to know or care that it's slop.
Very disheartening article on this: https://fortune.com/2025/12/30... [fortune.com] . There's no way YouTube will pull the plug on this. If it's making this kid $700,000 a year, YouTube themselves are making far more, despite having to store all that slop.
Re: (Score:3)
Re: (Score:2)
Doesn't seem to matter to them as along as the money rolls in, which it is.
Too many AI slop indeed (Score:3)
One time I got 'fed' a series videos of basically babies being in precarious (dangerous) situations, only to be saved by a cat (with the parents' voices in the background calling the cat 'good' for saving the baby - yeah, that absurd). They were all obviously AI generated (like how the baby 'about to fall off the stairs' went physically through the railing posts somehow, to say nothing of the weirdness of the stairs themselves).
Of course there were some idiots that actually believed the obvious fake videos (if they weren't fake comments themselves), but many more flagging it as obvious "AI slop", with the video poster replying to those with "Are you not amused?".
I "promptly" reported the videos as 'child abuse'.
The one good possibility with AI slop (Score:2)
In Vernor Vinge's "Rainbows End" (scifi written in the '00s but takes place approximately around now) there is a thing called "Friends of Privacy" which makes it hard to find relevant information about a person, by overwhelming the truth with a sea of spam/misinformation. If you google your ex-wife and get 200 different possible addresses (all of them credible), then you don't have her address.
I thought that was unrealistic and wouldn't really happen, of course it's starting to happen.
But it's happening (on
They don't get it (Score:3)
He says they are going to combat AI Slop, in the same breath as he says they are rolling out tools to make creating it easier.
I don't think this guy understands what most people think "AI Slop" even is. "Remixing existing content", *IS* slop. It is low effort, low value, garbage.
But it's all low quality. (Score:2)
AI slop (Score:2)
Resurrecting the Dead (Score:2)
I've seen a recent flood of AI slop channels that use the voices of long dead people like Richard Feynman and Christopher Hitchens to present political news (Hitchens) and science shows (Feynman). The voices are so close to the originals that it's easy to confuse historical content with this slop.
Who chooses? (Score:2)
Who gets to choose what is considered AI Slop or not? My kids like to watch these videos of a guy telling stories and to aid in his story telling he has little short segments of videos sometimes with AI generated content (@ElliotSimms) is that AI Slop?
\o/ (Score:1)
Allowing assholes to completely clone other human beings and misrepresent their content as content created by the target is reprehensible. A tiny unclear note that 'This is AI slop' does not resolve the problem. If it's AI Slop, that should be RIGHT IN THE MIDDLE OF THE THUMBNAIL used to gain clicks. Everyone involved is an asshole:
* If you have something to say - say it - YOURSELF.
big pain (Score:1)
it is a big pain, tons of video with a subject made so you clic and that just contain a mix of correct and incorrect informations (and which dont clearly say it is AI generated)
and youtube report button doesnt even have a AI generated/clicbate choice
if they dont act on it, it will kill human content creators as they disappear among those videos and they will not earn enough money to create real content.
Suck it YT (Score:2)
I saved a YT playlist to my YT over a decade ago and just received a "strike" for having something in that video playlist that was sexual. The playlist had hundreds of videos and I didn't watch any of them past the first few. I tried to appeal the strike but that was denied as well. Youtube can burn in hell at this point, I'm canceling youtube premium.