Giving Credit Where Credit Is Due: News Outlet Sues AI for Removing Copyright Information

DEFENDANTS’ INTENTIONAL REMOVAL OF COPYRIGHT MANAGEMENT INFORMATION FROM PLAINTIFF’S WORKS IN THEIR TRAINING SETS

96. ChatGPT and Copilot do not have any independent knowledge of the information provided in their responses. Rather, to service Defendants’ paying customers, ChatGPT and Copilot instead repackage, among other material, the copyrighted journalism work product developed by Plaintiff, and others, at often considerable expense.

97. When providing responses, ChatGPT and Copilot give the impression that they are an all-knowing, “intelligent” source of the information being provided, when in reality, the responses are frequently based on copyrighted works of journalism that ChatGPT and Copilot simply mimic.

98. If ChatGPT and Copilot were trained on works of journalism that included the original author, title, copyright notice, and terms of use information, they would have learned to communicate that information when providing responses to users unless Defendants trained them otherwise.

99. Based on the information described above, thousands of Plaintiff’s copyrighted works were included in Defendants’ training sets without the author, title, copyright notice, and terms of use information that Plaintiff conveyed in publishing them.

100. Based on the information above, including the OpenAI Defendants’ admission to using the Dragnet and Newspaper extraction methods, which remove author, title, copyright notice, and terms of use information from copyright-protected news articles published online, the OpenAI Defendants intentionally removed author, title, copyright notice, and terms of use information from Plaintiff’s copyrighted works in creating ChatGPT training sets.