➤ Discovery Battles Over AI Training Data: As AI-related copyright and privacy cases move into discovery, plaintiffs are now seeking access to the datasets used to train large AI models. Defendants are objecting on the basis of burden, confidentiality, and proprietary-information grounds. A recent filing in In re Google Generative AI Copyright Litigation illustrates the growing disputes over whether, and to what extent, training data must be produced.

➤ Discovery Into AI-Driven Claims Handling: Discovery is also underway in the health-insurance cases involving alleged algorithmic claim denials. The Rule 26(f) Joint Status Report and Discovery Plan in Barrows v. Humana provides a detailed look at the categories of information plaintiffs are now seeking, including internal documentation of AI-based medical-necessity tools and the extent of human oversight in claim decisions.

➤ Deloitte to Issue Refunds After AI-Generated Errors: Deloitte has agreed to reimburse the Australian government after submitting financial reports that contained AI-generated hallucinations. The incident has drawn attention to the risks of relying on generative AI for professional or technical analysis without adequate human review.

➤ Judicial Use of AI Under Scrutiny: Senator Chuck Grassley, Chair of the Senate Judiciary Committee, sent letters to two federal judges - Judge Henry Wingate (S.D. Miss.) and Judge Julien Neals (D.N.J.) - seeking explanations for apparent AI-generated hallucinations that appeared in recent judicial orders. The letters reflect growing concern about the use of AI tools by court personnel and the need for clear standards to prevent incorrect or fabricated citations from entering the judicial record.