YouTube has announced an important change in its transparency policy. And starting this month, the platform will stop depending solely on the boxes referring to AI that content creators check when uploading their videos and will begin to detect and label the videos generated with this technology itself. In addition, it will make these labels more visible both in the usual videos and in the Shorts. We tell you all the details.
What exactly changes. Until now, the responsibility fell on whoever uploaded the video. Starting in 2024, YouTube requires creators to disclose when they use realistic AI that could be mistaken for a real person, place, or event. The problem is that the creators didn’t have much incentive to be honest.
From now on, according to explains According to the company itself, if a creator does not indicate that they have used AI but their systems detect “significant use of photorealistic AI”, the label will still be applied automatically.
More visible labels. The other big change is where those notices appear. Previously, the information was hidden in the extended description of the video, in a dedicated section, so that only those who specifically went to look for it saw it. For long videos, the label (a symbol with the word “AI” next to an information icon) will now appear just below the player, above the description.
In the Shorts it will be shown superimposed over the video itself. YouTube affirms that with this change the viewer gets the context “at a glance.”
What the system does not cover. The featured tag is intended only for photorealistic or substantially AI-modified content. Videos that are animated, clearly unrealistic, or have minor tweaks will continue to display the warning only in the extended description. Here YouTube is clearly differentiating between AI that can confuse the user or create disinformation and AI for entertainment purposes.
How YouTube detects AI. The truth is that the company has not been too specific about how it will detect AI in videos. Which yes mention There are two cases that activate the label without exception: videos with C2PA metadata indicating that they are completely generated by AI, and those created with Google’s own tools, such as Veo or Dream Screen.
On the other hand, if a user believes that their video has been flagged in error, they can correct the status in YouTube Studio, except in the two situations mentioned where the flag is permanent.
A reasonable decision. The move comes shortly after Google presented in your I/O conference the model family Gemini Omnicapable of generating video with AI in a realistic and increasingly precise way. Added to this is that YouTube and the rest of the social networks have been dealing with the avalanche of ‘AI Slop‘, especially in the Shorts. The company also claims to have improved its deepfake detection technology, which now allows any adult to search for matches of their face on the platform.
No impact on money or referrals. YouTube assures that carrying an AI tag will not affect how a video is recommended or its ability to monetize. The company maintains that the goal is to “balance transparency with creator control,” not to penalize the use of the technology.
And now what. It remains to be seen how reliable the automatic detection will be and how many videos will continue to escape the system, since YouTube itself admits that there may be AI content without the new label. The platform has been criticized for the inconsistency of its labeling system to date, so we will have to wait to find out if the system ends up working.
Cover image | YouTube
In Xataka | Nvidia CEO Jensen Huang: “People talk about AI eliminating jobs. It’s complete nonsense”
GIPHY App Key not set. Please check settings