The Honorable James A. Soto, United States District Judge in the District of Arizona, entered a unique GenAI standing order on the use of Generative AI for both research, which is common, and writing, which is less so. See REL v. ASSET LIVING, LLC, No. CV 23-348-TUC-JAS (D. Ariz. Aug. 4, 2023) (case dismissed on October 25, 2023). Judge Soto’s order on GenAI states that it applies to any case before him, although it appears it is not listed on the official website of U.S. District Court of Arizona, as a special order of Judge Soto, or included in the District Court’s local rules or general orders.
The order states:
(1) If any portion of a pleading or other document filed on this Court’s docket has been drafted (in whole or in part) using generative artificial intelligence, including, but not limited to ChatGPT, Harvey.AI, or Google Bard, all attorneys and pro se litigants filing such pleadings or other documents shall verify that any language that was generated in any form by AI was checked for accuracy by using print reporters, traditional legal databases, or other reliable means by a human being.
Soto goes on to observe:
AI systems hold no allegiance to any client, the rule of law, factual or legal accuracy, or the laws and Constitution of the United States, and are not factually or legally trustworthy sources without verification by a human being.
Judge Soto continues with a detailed and unique requirement for filing a separate Notice of Use of Artificial Intelligence, which discloses all language originally produced by generative AI.
(2) If any portion of a pleading or any other document filed on this Court’s docket has been drafted (in whole or in part) using AI, all attorneys and pro se litigants filing such pleadings or other documents shall also simultaneously file a separate document entitled “Notice of Use of Artificial Intelligence” that: (a) identifies the AI program used and specifically identify all specific portions (i.e., citing page and line numbers) that have been drafted (in whole or in part) by AI[1]; and (b) certify that any language that was generated in any form by AI was verified for accuracy by using print reporters, traditional legal databases, or other reliable means by a human being.
As the endnote points out this requires the disclosure of particular page and line numbers.
