Sarah: Here are today’s top five eDiscovery AI stories for October 14, brought to you by eDiscovery Decoder.
George: First, EDRM reports that the Montana Commission on Rules of Evidence amended its proposed Rule 901 language for deepfakes after public comments. The revision is meant to avoid confusion when a judge, rather than a jury, is the factfinder. [source]
Sarah: As EDRM describes the proposal, if a challenger makes a sufficient showing that a reasonable factfinder could find generative-AI fabrication, the proponent would need to show that the item is more likely than not authentic. It remains a proposed rule, not a final rule. [source]
George: Second, a JD Supra article says AI-related discovery continues to turn on familiar relevance and proportionality principles. Its examples distinguish disputes where AI use is central from matters where prompts or outputs may bear more narrowly on intent, knowledge, or another contested issue. [source]
Sarah: The authors also stress fit and burden. They cite Warner versus Gilbarco, where a broad demand concerning third-party AI tools was found irrelevant or, at most, marginally relevant and disproportionate. Their larger point is a developing framework, not a rule that every AI interaction is discoverable. [source]
George: Third, eDiscovery Today reports on Villanueva versus Las Vegas Metropolitan Police Department. A Nevada district judge held that requiring production of an expert's entire, roughly forty-page ChatGPT history was clearly erroneous or contrary to law, while determining that some material still had to be disclosed. [source]
Sarah: The reported distinction depended on use. AI-generated case summaries considered in forming the expert's opinion qualified for disclosure as facts or data. His questions and prompts, deposition-preparation material, and draft-report material were protected because they implicated mental impressions, trial preparation, or expert-draft protection. The ruling was therefore not all-or-nothing.
An automated news discussion, not legal advice. Consult the original reporting for context.