Hot Potatoes: When it comes to high-tech companies that train AI models, everything seems to be a fair game. For example, Google has trained Gemini and Veo 3 using some of YouTube's billions of videos, and many creators don't realize that's happening.
With over 20 billion videos on the platform, YouTube is a treasure trove for AI companies to utilize and is already owned by many.
YouTube owner Google uses content to train AI models, CNBC reports. The company later confirmed that it was doing this, but it only uses a subset of the video and respects specific contracts with creators and media companies.
“We have always used YouTube content to improve our products, and this has not changed with the advent of AI,” a YouTube spokesperson said in a statement.
https://www.youtube.com/watch?v=od9lve4svqi
YouTube has acknowledged the need for safeguards in this area, and is investing in protection so creators can protect their image and likeness.
However, many experts point out that most creators and companies don't know that Google trains models with content. There is also no way for people to opt out of making their work use this way.
The report means that the size of YouTube's video library means that even if only 1% of the video is used for training purposes, it is equivalent to 2.3 billion minutes of content.
The situation has become more relevant since Google unveiled the VEO 3 video model, which allows you to create very realistic video clips. Like many industries, the irony is that the content people create is being used to train AIs that can ultimately replace them, or at least impact revenue in a competitive market.
https://www.youtube.com/watch?v=2cqurqidzx8
Some creators take a different perspective. They plan to use VEO 3 to create content, even if they are trained with their own work.
There have been cases where other companies use YouTube to train AIS without the creator's knowledge. It was reported last year that Openai had transcripted more than a million hours of YouTube videos to train LLMS. Nvidia did the same thing, but at one point she was rubbing down 80 years of videos every day. The company argued that this lies in the “spirit of copyright law.” Humanity, Apple, and Salesforce also relied on YouTube for AI training data.
Google now allows creators to opt out of third-party training from AI companies such as Amazon and Nvidia, but there is no option to stop Google from doing the same thing.
Image credits: Jordan Gonzalez
