The allegation
The New York Times and fellow publishers asked the court for sanctions, alleging OpenAI concealed internal tools that identify copyrighted journalism in ChatGPT outputs. Depositions surfaced 78 million de-identified chat logs and internal searches of training data. These are allegations, not verdicts. But transparency in discovery is a promise a company makes to society, not just to a judge.[1], [2]
The lesson
Media and image models are built on the labor of journalists and artists; these cases will decide who pays that bill. My assessment is straightforward: every input taken without asking the creator is not a borrowing, it's a debt. And debts get paid — at the courthouse, if nowhere else.[1]