News publishers complain OpenAI uses articles to train ChatGPT
News publishers complain OpenAI uses articles to train ChatGPT
Occurred: February 2023-
Report incident 🔥 | Improve page 💁 | Access database 🔢
Prominent news publishers accused OpenAI of using their articles without their knowledge or consent to train its AI models and systems, including ChatGPT.
CNN, Reuters, The News York Times, the Wall Street Journal, and other publishers accused AI companies and developers, including OpenAI, of using their articles to train their AI and machine learning software in violation of their terms of use.
In October 2023, News/Media Alliance, an US-based initiative representing over 2,200 publishers, argued that AI firms regularly used the information in news stories without authorisation, and violate laws protecting intellectual property in a white paper and comments to the US Copyright Office.
The group’s research claimed datasets, such as the C4 language dataset, 'significantly' overweighted content from news, magazines and digital media sources, using it 5 to almost 100 times as frequently as other content.
It also suggested chatbots copy and use publisher content in their outputs to users, putting them in competition with news outlets.
Operator: ABC News; Axios; BBC; Bloomberg; CNN; Condé Nast; Daily Mail & General Trust; Dow Jones; Disney; ESPN; Guardian News & Media; Hearst; Insider; The Atlantic; The New York Times; Reuters; Vox Media
Developer: OpenAI
Country: USA; UK
Sector: Media/entertainment/sports/arts
Purpose: Generate text
Technology: Chatbot; Generative AI; Machine learning
Issue: Copyright; Transparency
News/Media Alliance (2023). Submission to US Copyright Office
News/Media Alliance (2023). How the Pervasive Copying of Expressive Works to Train and Fuel Generative Artificial Intelligence Systems Is Copyright Infringement And Not a Fair Use (pdf)
Page info
Type: Incident
Published: December 2023