OpenAI Accuses Plaintiffs' Lawyers of Funding and Laundering Copyright Evidence

OpenAI and Microsoft say plaintiffs’ lawyers secretly funded and then smuggled questionable research into a consolidated copyright lawsuit to manufacture the missing “market dilution” evidence judges had demanded. After conflicting California rulings on whether training is fair use left judges asking for concrete proof that generative models flood markets, a series of preprints by Tuhin Chakrabarty, Jane Ginsburg and coauthors claimed AI-generated books dilute sales and that fine-tuning makes models verbatim-recall copyrighted works. Those studies drew enthusiastic media coverage despite methodological flaws flagged by critics: reliance on Kindle Unlimited data that mismeasures author income, apparent author bias (Chakrabarty dismissing critique on moral grounds), and expert rebukes (Yale’s A. Feder Cooper found the memorization claims overstated and poorly reported).

The filing accuses Susman Godfrey of paying Chakrabarty to produce the market-dilution study, failing to disclose that funding in his disclosed resume, and then “laundering” the work into the case outside the normal expert-report and discovery processes so defendants couldn’t properly challenge it. The complaint notes the paper was self‑published, the paid researcher was listed as a potential but unused expert, and disclosures were repeatedly incomplete. If true, the conduct would undercut a central pillar of plaintiffs’ evidence and could amount to a serious procedural fraud on the court.