I know there’s a lot of controversy over AI and the use of copyrighted material in the training of AI models, and it got me to thinking - what content was used to train the models we’re using in Video AI? Was copyrighted material involved or was everything either public domain or owned/licensed by Topaz?
From what I know, none of that matters since the output of TVAI, is the input. There is zero percent chance that TVAI will start to output copyrighted works that it was trained on.
I know that what Topaz’s tools do is a completely different animal from more controversial Generative AI, and none of that training material appears in our outputs, but I’d argue that the content used to train the models still matters regardless. If theoretically a model was trained on material that wasn’t cleared, that would present some clear ethical concerns in my view.
All of the content used to train the AI models has been cleared and licensed for use in the training. The video files used were sourced for their specific resolutions or conditions due to the AI models’ needs and the model’s desired function.
1 Like
Wonderful! Thanks for confirming!
Does this apply to your image datasets, as well, or just video?
You mention plenty that Gigapixel is trained using “millions of images” but I can’t find anywhere that you specifically certify how those images were obtained and whether they are licensed or already owned by yourselves.
Even if the final product technically qualifies as the original work of the owner of the original image, it’s important to me to know that the training itself was ethically sourced before investing in something like this…
Thank you for your time!
Yes, this applies to our AI models for the image apps as well. Our research team used in house source material and licensed other material as needed to train all of our models.
1 Like
