Researchers Accuse OpenAI of Using Paywalled O'Reilly Books for Training

OpenAI recently announced the launch of 'Academy', a platform designed to enhance AI literacy through free online and in-person courses, developed in collaboration with educational institutions. The initiative aims to provide foundational knowledge and specialized training on AI technologies. However, OpenAI is facing serious allegations regarding its GPT-4o model, which is accused of being trained on copyrighted content from O'Reilly Media without permission. Reports from the AI Disclosures Project indicate that GPT-4o shows a significantly higher recognition of this paywalled content compared to previous models. This controversy raises important questions about the ethics of data sourcing in AI training and the implications for copyright. Critics caution that relying on unlicensed material could undermine trust in AI technologies and emphasize the need for transparent data practices.