Kaat Scheerlinck’s Post

News Update U.S. Lawmaker Introduces Bill to Create AI Transparency On April 9, U.S. Representative Adam Schiff of California introduced a bill titled the Generative AI Copyright Disclosure Act of 2024. The main purpose of the proposed legislation is to require entities that create or significantly alter training datasets used for building generative AI systems to submit a notice to the Register of Copyrights, providing details about any copyrighted works included in those datasets. #ai #genai #news https://lnkd.in/dY6ZKNEV

Builders of generative AI tools have argued that prompts do not reproduce the training data, which should protect them from claims of copyright violation. Some audit studies have shown, though, that end users of generative AI can issue prompts that result in copyright violations by producing works that closely resemble copyright-protected content.

Great take by David Tollen on the current discussion around generative #AI and #copyright infringement. Here's the heuristic: think about the AI model and the training data as though it was your own mind. Publicly available information is free for you to read and so shouldn't be infringing any copyright if used to train the AI model. If you have to pay to read someone's content, then so should the AI. Infringement doesn't happen on the way *in*, it happens on the way *out*. Read all you want and use it to make as many new ideas as you like, but don't reproduce it word-for-word and call it your own. Same rule should hold true for AI outputs. Meghan Anzelc, Ph.D. I'd be interested to hear your thoughts on this topic as well. https://lnkd.in/eUZAuDqF

Would transparency be the right solution for a harmonic intersection between AI and Copyright? In this new article, Vincent Fauchoux and I discuss the delicate issue of transparency in the AI Act. #copyright #ai #aiact

🤖📚 Exploring the copyright risks of Generative AI (GenAI). How do we safeguard intellectual property in the age of AI-generated content? Discover key issues and strategies to mitigate copyright challenges. #GenAI #CopyrightProtection #AI https://lnkd.in/dPdZ7meu

AI chatbots are becoming increasingly sophisticated, and the question of who owns the copyright to the data they are trained on is a complex one. The article explores this issue, highlighting a lawsuit between the New York Times and OpenAI/Microsoft for $450 billion in damages, alleging the unauthorized use of NYT articles to train ChatGPT. 🤯 On one hand, AI companies argue that using massive amounts of text and code to train chatbots is a fair use of copyrighted material. On the other hand, news organizations and other copyright holders argue that AI chatbots that compete with them should share their profits. This is a fascinating and important legal battle with far-reaching implications. Should AI chatbot data be copyrighted? What are your thoughts? 🤔 #AI #copyright #chatbots #futureofwork https://lnkd.in/egCVNJUM

AI chatbots are becoming increasingly sophisticated, and the question of who owns the copyright to the data they are trained on is a complex one. The article explores this issue, highlighting a lawsuit between the New York Times and OpenAI/Microsoft for $450 billion in damages, alleging the unauthorized use of NYT articles to train ChatGPT. 🤯 On one hand, AI companies argue that using massive amounts of text and code to train chatbots is a fair use of copyrighted material. On the other hand, news organizations and other copyright holders argue that AI chatbots that compete with them should share their profits. This is a fascinating and important legal battle with far-reaching implications. Should AI chatbot data be copyrighted? What are your thoughts? 🤔 #AI #copyright #chatbots #futureofwork https://lnkd.in/e2ddKeqy

Data used to train AI, particularly generative AI, is often sourced from the internet using web scraping tools. As one can imagine, a lot of this data will be owned by someone or something and protected via copyright. While the use of this data could lead to intellectual property infringement, in the UK, there are some exemptions available which permit the use of a copyright work or database, though there are strict criteria to meet. Our article provides an overview of these exceptions, as well as the current law in the UK on use of copyright works to train AI. #AshfordsUnlockingAI #DataScraping #AI #TrainingAI #IntellectualProperty https://lnkd.in/eBY-VEkN

AI chatbots are becoming increasingly sophisticated, and the question of who owns the copyright to the data they are trained on is a complex one. The article explores this issue, highlighting a lawsuit between the New York Times and OpenAI/Microsoft for $450 billion in damages, alleging the unauthorized use of NYT articles to train ChatGPT. 🤯 On one hand, AI companies argue that using massive amounts of text and code to train chatbots is a fair use of copyrighted material. On the other hand, news organizations and other copyright holders argue that AI chatbots that compete with them should share their profits. This is a fascinating and important legal battle with far-reaching implications. Should AI chatbot data be copyrighted? What are your thoughts? 🤔 #AI #copyright #chatbots #futureofwork https://lnkd.in/eDgCg_HP

One problem is that output from an AI tool can be very similar to the underlying copyright-protected materials it was trained on, such as books, articles, or artworks. Leaving aside how generative models are trained, the challenge that the widespread use of generative AI poses is how individuals and companies could be held liable when generative AI outputs infringe on copyright protections.