en.Wedoany.com Reported - YouTube announced on Wednesday (27th) that it will strengthen the detection of artificial intelligence-generated videos posted by users and automatically add labels to highly realistic content to indicate the use of AI technology.
The rules implemented by the company since 2024 primarily rely on users self-disclosing information, but this requirement has not been fully enforced. For more realistic content, the platform will place warnings about AI-generated material in more prominent positions than before. For content made with AI but deemed by the company as "unrealistic, animated, or slightly modified," the label will remain in the video description field, with lower visibility.
These updates were announced during interviews with journalists from Brazil and European countries. In Brazil, these rules directly correspond to regulations from the Superior Electoral Court (TSE) regarding electoral propaganda and political content on social media. In 2024, the court had already mandated the labeling of synthetic content in electoral propaganda. This year, joint liability was also added for social media platforms that fail to remove content deemed a "risk." One such risk is publishing content generated or modified by AI that does not comply with labeling rules and other electoral prohibitions. The TSE also maintains a ban on deepfakes, regardless of whether the intent is to harm or benefit a candidate, and prohibits the posting of AI-related content within 72 hours before an election.
In 2024, YouTube stipulated that channel owners must state whether AI tools were used when uploading videos. Nearly two years later, the company announced that starting this month, it will implement "new internal signals to help identify AI-generated content," and if the platform's systems detect "significant use of realistic AI," a label will be automatically applied. The announcement stated that as the technology continues to improve, users can change this automatic classification. According to the company, the only situation where this status remains permanent is when content is created using YouTube's own AI tools, or the video file carries "C2PA metadata" (a type of technical information equivalent to a seal indicating the content was entirely AI-generated).
When asked whether a user's status change is necessarily accepted or still requires analysis by the YouTube team, the company only responded that this update made by the user "is not an appeal." This implies that for cases not falling under the permanent status, there is no additional analysis. According to YouTube, the label itself will not change how the video is recommended on the platform, nor will it lead to demonetization. The company also stated it reserves the "right to take action" against users who repeatedly fail to disclose AI use.
On the other hand, it remains unclear to what extent the company has already adopted this automatic label application. The 2024 announcement already claimed it could be done in certain cases, "especially if the modified or artificial content has the potential to confuse or mislead people." When asked whether labels have been automatically applied and under what circumstances, the company replied: "We have long invested in systems that can apply our policies, and this update is specifically designed to leverage the power of automated detection to help creators disclose information and provide viewers with the transparency they have been asking for."
The company also announced it is expanding a deepfake detection tool based on individuals' facial appearances to all users aged 18 and over. This tool will track videos already published on the platform where their face appears and may have been modified or created by AI, rolling out gradually in the coming weeks. Through the platform's initial scan, users can view the listed content and request the removal of parts generated or modified by AI. According to the company, removal is not automatic and will depend on analysis and criteria within its privacy guidelines.
Amjad Hanif, YouTube's Vice President of Creator Products, stated: "What makes this tool unique in the industry is that it will search all content published on YouTube to find instances where [your face] matches content created without your knowledge." According to Folha de S.Paulo, this measure began piloting in March this year for public figures such as politicians and journalists. Users who consent to register need to submit official identification and record a short facial video (like a selfie). Users must have a YouTube channel (even if unused) to use the tool. According to the rules, factors the platform considers when evaluating removal requests include: whether the content is synthetic or modified, whether it is realistic, or whether it presents parody, satire, or other public interest aspects.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
