The announced this Wednesday (27) that it should increase the content and that, in the case of realistic content, it will automatically apply labels identifying .
Until then, the company’s main rule, in force since 2024, was based on information provided by the user, which is not always done.
For this type of more realistic content, the platform also stated that it will position the warning that it is material produced with AI in a more prominent place than it has done so far. For content made with AI, but considered by the company as “unrealistic, animated or slightly altered”, this label will remain in the video description field and, therefore, with less visibility.
The updates were announced in an interview with journalists from Brazil and European countries.
In the Brazilian scenario, these rules dialogue directly with norms established by the (Superior Electoral Court).
In 2024, the court had already predicted that the labeling of synthetic content would be mandatory in electoral propaganda. This year, joint liability for social networks was also included if they do not remove content considered “at risk”.
One of these risks would be the dissemination of content generated or altered by AI that does not comply with labeling rules and other electoral prohibitions.
Among other items on the topic, the TSE also maintained the ban on deepfake, whether to harm or favor candidates, and included a ban on the publication of content with AI from 72 hours before the election.
In 2024, YouTube established a rule that channel owners must inform, when publishing a video, whether they used AI tools. Now, nearly two years later, the company is announcing that, starting this month, it will implement “new internal signals to help identify AI-generated content” and that if the platform’s systems detect “significant use of photorealistic AI,” a label will automatically be applied.
The announcement further states that “as this continues to improve,” users will be able to change this automatic rating. The only cases in which this status will be permanent, according to the company, are for content created using YouTube’s own AI tools or when the video file has “”, a type of technical information that works as a stamp that that content was entirely generated by AI.
A Sheet questioned whether this change in status by the user is necessarily accepted or whether an analysis is still being carried out by the YouTube team. In response, the company only added that this user-made update is “not an appeal.” Therefore, implying that there is no additional analysis for situations that do not qualify as permanent.
According to YouTube, the label, in itself, does not change the way a video is recommended on the platform or mean that it can no longer be monetized. The company also says that it reserves “the right to take action” against users who repeatedly do not disclose the use of AI.
On the other hand, it is not clear to what extent the company eventually adopted this type of automatic label application. That’s because the 2024 announcement already stated that this could be done, in some cases, “especially if the altered or artificial content has the potential to confuse or mislead people.”
A Sheet asked the company whether the label was already being applied automatically and, if so, in what situations, but the response sent did not answer the question. “We have long invested in systems capable of enforcing our policies, and this update specifically aims to use the power of automated detection to help creators disclose information and provide viewers with the transparency they have been requesting,” the company said.
Deepfake Detection Tool
The company also announced that it is expanding access, to all users over the age of 18, of a tool that tracks, based on a person’s appearance, content published on the platform in which their face appears and has been potentially altered or created with.
This availability should occur gradually over the next few weeks.
From this initial scan carried out by the platform, the user will be able to review the content listed and request the removal of those generated or altered by AI. According to the company, the withdrawal will not be automatic and will depend on analysis and criteria defined in its privacy guidelines.
“What’s really unique about this tool in the industry is that it will search everything that has been published on YouTube to find if there are any matches [do seu rosto] for something that was created without your knowledge,” said Amjad Hanif, vice president of creator products at YouTube.
As shown by Sheetlike politicians and journalists in March this year.
Those who agree to sign up will need to submit an official ID, as well as record a short video of their face, such as a selfie. You must have a registered channel on YouTube —even if unused— to use the tool.
According to , among the factors considered by the platform to evaluate the removal request is whether the content is synthetic or altered, whether it is realistic, or whether it presents parody, satire or another aspect of public interest.
