Publishers suing OpenAI over copyright infringement say the company hid evidence that could have exposed how deeply ChatGPT draws on their work.
The New York Times and other news organizations filed a motion for sanctions against OpenAI, alleging the company concealed internal tools and datasets capable of identifying copyrighted journalism in ChatGPT outputs. The publishers claim this evidence was withheld deliberately during discovery. The motion escalates what was already one of the highest-stakes copyright cases in AI history.
Sanctions motions are not filed lightly — they signal the publishers believe OpenAI's conduct crossed from aggressive litigation strategy into misconduct. If a judge agrees, consequences could range from fines to adverse inference instructions, meaning a jury could be told to assume the hidden evidence was damaging to OpenAI.
The case has always been about more than back pay for old articles. It is a referendum on whether large language models can be trained on copyrighted text without a license. OpenAI's ability to hide the seams between training data and outputs is precisely what the publishers are trying to expose — and what OpenAI, apparently, would prefer remain unexposed.