NVIDIA faces scrutiny over alleged unlicensed data scraping for AI models
The post NVIDIA faces scrutiny over alleged unlicensed data scraping for AI models appeared on BitcoinEthereumNews.com.
Leaked documents obtained by 404 Media suggest NVIDIA engaged in unlicensed data scraping, using movie and game footage from across the internet to train its artificial intelligence products. The leaked documents reveal that they were trying to download full movies from various channels, including Netflix, and their primary interest was in YouTube videos. From the emails obtained by 404 Media, the project managers intended to employ between 20 and 30 virtual machines on Amazon Web Services to obtain 80 years of videos in a day. NVIDIA defends its actions and invokes fair use provisions Data scraping is the practice of extracting video, textual, and audio content from the internet without the permission of the content owners to train AI models. This practice could be seen as the use of content from social media platforms that contain copyrighted content. NVIDIA has said that it did not break any copyright laws in the process of data scraping. The company also stated that its activities fall under the fair use doctrine because it utilizes copyrighted material for training AI. Documents obtained from internal communications by 404 Media indicate that some NVIDIA employees expressed concerns over these data scraping activities. However, project managers allegedly downplayed the concerns, stating that legal concerns, for example, violations of YouTube’s Terms of Service, would be dealt with later on. One employee pointed out that NVIDIA’s AI engineers tried to get as many game clips as possible to enrich the training corpus. This entailed streaming the gameplay to NVIDIA’s GeForceNow cloud service to record gameplay videos in high definition.Jim Fan, senior research analyst, in internal messages also stressed the importance of such footage as the input for the training of the AI model. Company takes steps to manage public perception of data practices The documents also detail NVIDIA’s attempts…
Filed under: News - @ August 5, 2024 9:16 pm