4 items across 4 digests
Human Archive, founded by UC Berkeley and Stanford researchers, pays Indian gig workers to wear camera-equipped caps and sensors to collect real-world training data for AI and robotics labs. This approach addresses the critical need for diverse, real-world data to train more capable AI systems and robots.
Patreon CEO Jack Conte argues AI companies should compensate creators for training data, calling fair use arguments invalid. He points to AI companies licensing content from major publishers as evidence against their fair use defense.
Encyclopedia Britannica and Merriam-Webster are suing OpenAI for allegedly using nearly 100,000 copyrighted articles without permission to train their large language models. This lawsuit adds to growing legal challenges around AI training data and intellectual property rights.
Musk failed to block California's data disclosure law requiring AI companies to reveal training data sources, with the judge ruling public interest outweighs privacy concerns. This sets precedent for AI transparency regulations that could impact xAI and other AI companies' operations.
DailySand