A hundred examples of verbatim output
On 27 December 2023 The New York Times Company sued Microsoft and eight OpenAI entities, case 1:23-cv-11195. Exhibit J to the complaint runs to 127 pages: exactly one hundred numbered examples, each setting a Times article beside the model's output with the overlaps marked. The prayer for relief asks the court to order destruction under 17 U.S.C. section 503(b) of all GPT models and training sets that incorporate the paper's work.
Why it matters
Earlier suits argued that works had gone into training. This one argues something else and harder: that they stayed inside and can be pulled back out almost word for word. And for the first time the remedy sought is not compensation but the destruction of the model itself.
Exhibit J was filed the same day as document 1-68. Its example numbers run from one to a hundred and go no higher, so the count is exactly a hundred; the report this record was built from said more than 100. The complaint's own text gives no number, saying scores of examples. The suit seeks billions of dollars in statutory and actual damages. What the record does not claim. Nothing is said about the later course of the case or whether the destruction remedy was granted.