Serving technology enthusiasts for more than 25 years. TechSpot is the place to go for tech advice and analysis you can trust.
It seems that everything is fair play when it comes to training AI models by tech companies. Google, for instance, uses some of YouTube’s billions of videos to train Gemini 3 and Veo 3. Many creators are unaware of this. YouTube has more than 20 billion videos, which is a wealth of data that AI companies can use – many have already done so.
YouTube’s owner Google is using the content for its AI models. CNBCLater, the company confirmed that they do this but only use a subset and honor specific agreements with media companies and creators. In a statement, a YouTube spokesperson said
“We’ve always used YouTube content to make our products better, and this hasn’t changed with the advent of AI,” .
YouTube acknowledged that there was a requirement for safeguards in the area and has invested in protecting creators’ images and likenesses.
However, many experts have pointed out that the majority of creators and companies do not know that Google trains its models using their content. It’s impossible for anyone to opt out of this use.
According to the report, the size of YouTube’s video library means even if only 1% of videos are used for AI training, that is 2.3 billion minutes worth of content. This is 40 times more than the training data that competing AI models use, according to experts.
This situation has become even more relevant after Google announced its Veo 3 model, which can create video clips that are incredibly realistic. Ironically, as with many industries, content created by people is used to train AI that may eventually replace them or at least affect their income in a competitive market.
Other creators have a different view. They use or plan to use Veo 3 for creating content, even though it was trained on their original work.
Other companies have used YouTube to train their AIs, without the creators’ knowledge. OpenAI was reported to have transcribed more than a million YouTube videos in order to train their LLMs. Nvidia also did this, and at one time was scraping 80-years of videos every day – the company claimed that this was in “the spirit of copyright law.” Anthropic Apple and Salesforce turned to YouTube for AI training data.
Google allows creators to opt-out of third-party AI training from companies such as Amazon or Nvidia. However, there is no option to prevent Google from doing so.
Image credit: Jordan Gonzalez
