The string of losses OpenAI has suffered recently in the MDL litigation should give OpenAI pause.
For a brief summary of some of these losses:
What does ChatGPT think now of OpenAI’s chances in the litigation?
At least at the moment, it’s hard for OpenAI to be optimistic about the MDL litigation. There are a dozen or so lawsuits. Any one of them could result in substantial liability. And an adverse result in any combination of them, if not all of them, could be massive in terms of statutory damages. Plus, Judge Stein and Magistrate Judge Wang both gave major losses to OpenAI just recently.
OpenAI is now defending against infringement claims based on three different theories: (1) alleged use of shadow libraries, (2) use in training models, and (3) outputs in summaries of books and news articles.
Even if OpenAI wins on (2) fair use in training its models, that would not necessarily mean a victory on (1) or (3). If Judge Stein follows Judge Alsup’s approach in Bartz v. Anthropic, (1) use of shadow libraries might be treated as a separate use from (2) training. Plus, Judge Alsup expressly stated that the case did not involve any alleged infringing outputs, but if the plaintiffs found such outputs, they could file another lawsuit.
We asked ChatGPT for its assessment. Things don’t look good for OpenAI right now:
