top of page
  • Writer's pictureSarah Ruivivar

ChatGPT Responses Are Not Derivative Work, says OpenAI

Updated: Oct 31, 2023

Did you hear the latest buzz in the AI world? OpenAI's ChatGPT is under fire!

A group of authors, including the likes of Sarah Silverman and Paul Tremblay, have raised their voices against the AI, alleging it was trained on pirated copies of their books. Quite the plot twist, right?

OpenAI, however, has responded to these allegations, asking a US district court in California to dismiss all but one claim. They argue that the authors have misunderstood the scope of copyright, overlooking the exceptions and limitations that allow room for innovative technologies like large language models.

OpenAI maintains that even if the authors' books were a part of ChatGPT's massive data set, using copyrighted materials in transformative ways does not violate copyright. They argue that their aim is not to profit off distributing copyrighted materials, but to teach their models the underlying rules of human language.

The authors, however, see it differently. They argue that every output from ChatGPT represents a derivative work, and thus infringes on copyright. OpenAI counters this, stating that not every output can be considered a derivative work under the authors' theory.

OpenAI has also dismissed claims that ChatGPT's training models violate the Digital Millennium Copyright Act (DMCA). The authors argue that any output that copies their work but excludes copyright-management information, such as author names or publication years, violates the DMCA.

The legal battle is ongoing, and it remains to be seen how the court will rule. But one thing's for sure - this case could have far-reaching implications for AI and copyright law. So, stay tuned for more updates on this intriguing saga!

Made with TRUST_AI - see the Charter:

17 views0 comments


bottom of page