Summary:
Le gouvernement britannique fait face à un revers concernant son plan sur les droits d’auteur liés à l’IA, tandis que Reddit poursuit Anthropic pour utilisation non autorisée de données. Une étude révèle que l’IA pourrait aider à dater les manuscrits des rouleaux de la mer Morte, et une montée des agents d’IA est observée en Chine malgré des défis réglementaires. Enfin, une interview avec Yoshua Bengio souligne l’importance de la sécurité dans le développement de l’IA.
Original Link:
Original Article:
AI News Roundup – UK government’s AI copyright plan suffers setback, Reddit sues Anthropic over unauthorized data use for AI training, AI helps unlock secrets of Dead Sea Scrolls, and more. To help you stay on top of the latest news, our AI practice group has compiled a roundup of the developments we are following.
– BBC News reports on the latest parliamentary setback experienced by the U.K. government’s plan to grant AI companies access to copyrighted works unless the creators opt out. The House of Lords has added several amendments to the Data (Use and Access) Bill that would introduce transparency requirements to the legal structure, permitting copyright holders to know how, when, and by whom their works are used. Similar amendments were proposed for the bill in the House of Commons, but were rejected. High-profile opponents include Elton John and Paul McCartney. Supporters claim that requiring companies to get permission to use copyrighted works would “kill the AI industry in this country.”
– The Wall Street Journal reports that Reddit is suing Anthropic, alleging unauthorized use of Reddit data to train AI chatbot Claude. Reddit has made content licensing agreements with Anthropic competitors OpenAI and Google and claimed that Anthropic’s web scraping bots accessed Reddit’s content even after being blocked.
– The Economist reports on a study that used AI to date the Dead Sea Scrolls, suggesting they are older than previous estimates. Researchers analyzed letter shapes and carbon-dated samples with an AI model called Enoch. This could aid in dating other ancient manuscripts.
– The MIT Technology Review highlights a boom of AI agents in China following the launch of Manus. These agents can perform tasks like answering emails and booking flights, but face regulatory challenges. A version of Manus aimed at the Chinese market is under development.
– The Financial Times interviewed Yoshua Bengio on AI safety, emphasizing the need for focus on safety as companies push for capability in AI. Bengio founded the non-profit LawZero to promote safer AI systems.
View source: [McDonnell Boehnen Hulbert & Berghoff LLP](https://www.mbhb.com/intelligence/snippets/ai-news-roundup-uk-governments-ai-copyright-plan-suffers-setback-reddit-sues-anthropic-over-unauthorized-data-use-for-ai-training-ai-helps-unlock-secrets-of-dead-sea-scrolls-and/).