TikTok Takes Initiative to Label AI-Generated Content

In a move to combat potential misuse of artificial intelligence (AI), TikTok announced plans to implement labels for videos and images uploaded to its platform that were created using AI tools. This initiative utilizes a digital watermark technology known as Content Credentials.

The decision comes amidst growing concerns from researchers regarding the potential manipulation of AI-generated content to disrupt upcoming US elections. Notably, TikTok was already part of a coalition of 20 tech companies that signed an accord earlier this year vowing to combat this issue.

Previously, TikTok focused on labeling AI content generated within the app itself. This latest development expands their efforts to encompass videos and images created using external AI tools.

“We have established policies that prohibit unlabeled, realistic AI content,” explained Adam Presser, TikTok’s Head of Operations and Trust and Safety. “If such content surfaces on our platform, it will be removed for violating our community guidelines.”

Content Credentials, the technology underpinning this initiative, was spearheaded by the Coalition for Content Provenance and Authenticity. This group, co-founded by industry giants like Adobe and Microsoft, offers an open-source solution for companies to implement.

Several major platforms, including YouTube (owned by Google) and Meta (owner of Facebook and Instagram), have already expressed their intention to adopt Content Credentials.

The effectiveness of this system hinges on collaboration. Both the creators of AI generation tools and the platforms where this content is shared need to agree to utilize this industry standard.

For instance, when someone generates an image using OpenAI’s Dall-E tool, OpenAI embeds a watermark within the image file. Additionally, data is added that allows for later verification of any alterations made to the image. If this marked image is then uploaded to TikTok, the platform will automatically recognize it as AI-generated and apply the appropriate label.

It’s important to note that TikTok, owned by the Chinese company ByteDance, faces a complex situation in the US market. The platform boasts 170 million US users, but a recent law has been passed requiring ByteDance to either divest TikTok or face a ban. Both TikTok and ByteDance have challenged this law in court, citing potential violations of the First Amendment.

While the legal battle continues, TikTok’s proactive approach to labeling AI-generated content demonstrates their commitment to responsible content moderation and fostering a trustworthy platform for their users.