Vimeo has announced the introduction of AI content labels, joining the ranks of TikTok, YouTube, and Meta in requiring creators to disclose when their content is generated or manipulated by AI. This move aims to ensure that viewers can distinguish between real and AI-generated content, addressing growing concerns about the potential for misinformation.
The updated terms of service and community guidelines mandate that creators label content created with AI tools if it depicts realistic scenarios or persons. This requirement does not extend to clearly unrealistic content, such as animations or videos with obvious visual effects. The focus is on preventing viewers from mistaking AI-generated content for real events or actions, especially in cases involving celebrities or altered footage of actual events.
A distinct label will now appear at the bottom of videos, indicating that the content creator has voluntarily disclosed the use of AI. Creators can select a checkbox to specify whether they used AI for audio, visuals, or both when uploading or editing a video. This labeling system applies to videos utilizing Vimeo’s own AI tools, such as those for editing out long pauses or disruptions in speech.
Future Plans for Automated Detection
While the current system relies on creators to disclose AI-generated content, Vimeo is developing an automated system to detect and label such content. CEO Philip Moyer highlighted this long-term goal in an official blog post, stating, “Our long-term goal is to develop automated labeling systems that can reliably detect AI-generated content, further enhancing transparency and reducing the burden on creators.”
In addition to implementing AI content labels, Vimeo has taken steps to protect user-generated content from being used to train generative AI models. Moyer has emphasized that Vimeo prohibits AI companies from training their models on videos hosted on the platform. This stance is in line with YouTube’s policies, where CEO Neal Mohan has also made it clear that using videos on the platform to train models, including OpenAI’s Sora, violates their terms of service.
Vimeo’s introduction of AI content labels represents a significant step towards enhancing transparency and protecting viewers from misinformation. By ensuring that AI-generated content is clearly labeled, Vimeo joins other major platforms in addressing the challenges posed by advancing AI technologies. As the company works towards developing automated detection systems, the burden on creators will be reduced, further streamlining the process of maintaining transparency in video content.
See also: Menlo Ventures and Anthropic Launch $100M AI Fund to Fuel Early-Stage Innovation
