Google Trains AI Video Generator on YouTube, Creators Unaware
Colaboradores
Google is reportedly using its vast YouTube video library, encompassing approximately 20 billion videos, to train its AI models, including Gemini and Veo 3. This practice, as reported by CNBC, involves utilizing portions of these videos, amounting to billions of minutes of content, for AI model training.
While Google claims to adhere to agreements with creators and media companies, many creators were unaware that their content was being used for this purpose. YouTube asserts that it has always utilized content to improve its products and has implemented measures to help creators manage their image usage in the AI era. However, creators lack the ability to prevent Google from using their videos for its own AI models.
Concerns Arise Over AI-Generated Video Similarity
Experts and creators are expressing concern about the implications of this practice. Tools like Trace ID, developed by Vermillio, can detect overlaps between AI-generated videos and original content. Trace ID has found that Veo 3 has produced videos remarkably similar to existing YouTube content. For example, a Veo 3 video closely resembled a video from creator Brodie Moss, achieving a similarity score of 71 overall and over 90 for the audio.
Creators Divided on AI Competition
Some creators welcome the competition presented by AI, while others feel that their work is being used unfairly, without proper credit, consent, or compensation. This news coincides with growing pushback from the entertainment industry, as exemplified by Disney (DIS) and Universal (CMCSA)'s recent copyright lawsuit against AI company Midjourney.
Google's Response and Legal Responsibility
Google states that it will assume legal responsibility if users encounter copyright complaints related to content generated with Veo 3. YouTube has also partnered with the Creative Artists Agency to assist top talent in managing their image usage in AI. However, some argue that YouTube's tools are unreliable. U.S. lawmakers, including Senator Josh Hawley, are also raising concerns about the ethical implications of AI training on copyrighted content.