Executives from Google and YouTube have stressed OpenAI scraping users’ data from its platforms for the purpose of AI training would be a violation of the company’s Terms of Service, The Verge reported.
OpenAI has been accused of transcribing huge volumes of video data from YouTube through its Whisper audio transcription model.
In a recent media interview, OpenAI’s CTO Mira Murati dodged answering questions on whether the company had used YouTube videos to train its newly released text-to-video model Sora.
Google spokesperson Matt Bryant noted seeing “unconfirmed reports” of OpenAI’s activity, reported The Verge, but company officials have not gone so far as to directly accuse OpenAI of harvesting its data.
(For top technology news of the day, subscribe to our tech newsletter Today’s Cache)
Google has also been hit with similar legal action and criticism over its AI offerings such as the Gemini model, and its alleged treatment of users’ data for AI training as well.
OpenAI previously defended its use of information as being protected by the fair use doctrine, but also pointed out its partnerships with various media outlets and claimed to obtain data from public sources as well.
Over the weekend, design and software company Adobe was hit with boycott calls by users who stood against the platform’s text-to-image generators. Those calling for a boycott claimed that such generative AI-enabled features represented the unoriginal use of many artists’ works without consent or compensation.
Published - April 08, 2024 09:18 am IST