On 18 May 2026, Magistrate Judge Thomas Farrish of the US District Court for the District of Connecticut ordered plaintiff Conservation Law Foundation to produce the generative AI prompts its expert witness, Dr Naomi Oreskes, used in preparing her expert report. This appears to be the first federal court decision requiring an expert witness to disclose AI prompts as part of discoverable methodology.
Dr Oreskes and her research assistant had used a commercially available generative AI tool (OpenAI’s GPT-4o, accessed through a private Microsoft Azure server) to filter and identify potentially relevant documents from the defendants’ discovery productions.
Why did the court treat prompts as methodology?
The court ruled the prompts are “part of the expert’s methodology” and therefore squarely within Rule 26(b)’s discovery scope. Methodology is not protected work product when it is the method by which the expert reached the opinions offered to the court. Prompts that select, filter, or frame source material sit inside that method.
The district court stayed the order pending resolution of the plaintiff’s objection. The stay does not erase the ruling’s significance. Counsel now have a federal decision that treats AI prompts as discoverable methodology, not as privileged scratch notes.
What should every expert who uses AI assume?
Assume the prompts will be requested. Design the workflow so that the prompts tell a story of professional rigour: defined scope, documented sources, and human review. An expert who cannot explain why a model was asked a particular question will struggle to defend the opinion that followed.
Check your governance exposure in three minutes.
This case is tracked in the Ecsper AI Risk Intelligence, which documents over 2,000 court cases involving AI-fabricated content.
