OpenAI Accused of Facilitating Copyright Infringement Through LLM Development and Training

Count III: Contributory Copyright Infringement by All Defendants

132. The above paragraphs are incorporated by reference into this Count.

133. In the alternative, to the extent a user may be liable as a direct infringer based on output of ChatGPT and/or Copilot, Defendants materially contributed to and directly assisted with the direct infringement by those users by jointly developing LLMs capable of distributing unlicensed copies and abridgements of the Registered Works, building and training LLMs using the Registered Works, and deciding what content is emitted by their products through the process of training them and developing them to conduct synthetic searching.

134. Defendants knew or had reason to know of the direct infringement by their users because Defendants undertake extensive efforts in developing, testing, or troubleshooting their models, (as to the OpenAI Defendants) have admitted that their products regurgitate material in response to user prompts, and have agreed to defend and indemnify certain of their users for copyright violations only when the users are using the products according to terms specified by Defendants.