A new $75 million lawsuit has been filed against Anthropic by authors who opted out of a previous settlement, focusing on how their books were used to train AI models. This legal action coincides with the publication of a scholarly paper proposing that AI model weights could be considered "probabilistic copies" of copyrighted material. The paper argues that the legal definition of a "copy" needs to evolve to encompass the way AI models store information, which could significantly impact the economics and compliance burdens for all AI training companies. AI
IMPACT This case and the 'probabilistic copy' theory could fundamentally alter how AI companies license training data and manage copyright compliance.
RANK_REASON A new lawsuit and a scholarly paper that could redefine copyright law for AI training data. [lever_c_demoted from significant: ic=1 ai=0.4]
Read on dev.to — Anthropic tag →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
