YouTube is cracking down on undisclosed AI videos, but won't penalize creators for making them

[Alfonso Maruccia]

In context: Generative AI is reshaping the content landscape, and YouTube is no exception. As AI generated videos become increasingly common on the platform, the company is updating how it discloses that content to viewers, making labels more visible and harder to overlook.

YouTube recently announced two changes aimed at improving how creators and viewers deal with generative AI output. The Alphabet-owned platform says its community expects full transparency around AI generated content, and is now rolling out a simplified labeling system for both long-form videos and Shorts.

Under the new system, an "AI" label is required whenever a video contains photorealistic content that has been meaningfully altered by AI or generated entirely by AI tools. For long-form videos, the label will appear above the description. For Shorts, it will display as an overlay directly on the video – both positions more prominent than before.

YouTube says the clearer labeling should help viewers quickly identify when a video relies heavily or exclusively on AI-generated content. The prominent label applies specifically to photorealistic and AI-altered material, while unrealistic, animated, or minimally altered videos will still carry an AI disclosure, though tucked within the expanded description.

The company is also introducing "internal signals" to automatically detect AI generated content and apply labels when creators have not done so themselves. The platform has been asking creators to label GenAI videos since 2024, and is now going the algorithmic way when a proper disclosure has not been provided. Creators who believe a video has been incorrectly flagged can dispute the label through YouTube Studio.

Needless to say, videos generated through YouTube's own AI tools such as Veo or Dream Screen will be permanently marked as AI-generated. The same applies to content carrying valid C2PA manifest metadata, a technical standard designed to reliably identify AI-generated media.

YouTube framed the changes as an effort to balance transparency with creator control, and was quick to clarify that the new labels will have no effect on how videos are recommended or monetized, which I think is a real shame considering how much AI-generated stuff now seems to be flooding the platform, particularly on long-tail searches.

The update is the latest in a series of AI-related moves from YouTube, whose grip on online video shows no signs of loosening. Google continues to make billions from the platform while pushing more aggressive ad formats on its users. In the meantime, the company has been steadily layering in AI features, while also using the platform's vast library of public videos to train its own language models and chatbot systems.

Permalink to story:

YouTube is cracking down on undisclosed AI videos, but won't penalize creators for making them

 

[The Talking Tech]

As much as I am against A.I taking over everything, it's ok for creators to have fun with their A.I creations, but as long as they don't go around deceiving viewers making them believe it's real life content and with platform holders heavily cracking down on those deceptions.

 

[ActualViking]

There is a AI blocking filter that helps avoiding AI slop when using Firefox for youtube watching. It's based on users reporting the channels with slop, so while not 100% it works well.

 

[RudyBob]

I don't think I much care if the video is an AI job. I don't like that so many voice overs sound British. They have been led to believe that if people hear a Brits voice whatever is being said has more gravitas.

 

[Aolish]

GOOD! This should have been done awhile ago. I'm tired of trying to guess if the video I'm watching is AI generated or not. I sometimes have to go through the comments as there are many who are better AI detectors than me. It seems that the AI slop is worse on animal videos. I love watching a lot of warm fuzzy animals get rescued and treated etc but with all the recent AI slop it's messed it up for me and now I only watch animal videos from official sources as I know those aren't AI generated.

 

[Megalomaniac]

I just got finished watching an AI video of Blade hunting down historical vampires.

And I really enjoyed "Superman Born on Viltrum".

I for one enjoy AI videos.

 

[Nobina]

Using AI to help you make a video is fine. Generating a whole video using AI is lazy and usually content without worthwile information. I can spot AI videos no problem, but it's a hassle searching through them to find what I want. If YouTube stops them making money from AI videos they will stop making them.

 

[PierreJG]

I personally don't like AI videos. The reason is that those people do not use their own creative skills. Anyone can use AI to make videos. Just my personal opinion. They should not be monetized. I watch a lot of videos that I know are not AI. AI makes a lot of mistakes, and it also kinda kill the purpose of using social media to create videos. Not sure if they are monetized, as I do not read rules too much. If they are, that will be unfair to the people who actually create legit videos. They will get the same money as a person who creates a proper video and will take them 5 minutes.

 

[Axeia]

How about an option to filter out AI generated content?
The amount of slop vs funny (haven't see any worthwhile use other than amusement when it comes to AI videos) when it comes to AI content is higher than I want to deal with.

For the few gems like when Trump tariffed the Heard and McDonald islands I'll find them through other means.