Generative Artificial Intelligence (AI) Use and Compliance Certification
(Federal)

Summary

This generative artificial intelligence (AI) use and compliance certification is a standard template attorneys can use to comply with judge-specific requirements concerning the use of generative AI apps, such as ChatGPT, Harvey.AI, or Google Bard, in drafting federal court filings. This template includes practical guidance and drafting notes. Given the recent explosion of generative AI tools available to and used by litigators, courts across the country have started evaluating the need for certain carefully enumerated requirements regulating the use of these tools in their proceedings. At least one federal court judge—Judge Brantley Starr of the U.S. District Court for the Northern District of Texas—has issued a standing order governing the use of generative AI to draft legal briefs. To guard against generative AI's potential for hallucinations, bias, and inaccurate or incorrect information, your judge may require you to certify either that: • No portion of any filing will be crafted by generative AI apps (e.g., ChatGPT, Harvey.AI, Google Bard) –or– • You will check any language drafted by generative AI for accuracy using print reporters or traditional legal databases See, e.g., U.S.D.J. Brantley Starr Specific Requirements, Mandatory Certification Regarding Generative Artificial Intelligence. In addition to any judge-specific requirements, you should also consider the following before using generative AI in your litigation: • Protect confidential information. Do not enter any information that is protected by the attorney client privilege or contains your client's confidential, sensitive, or propriety information. Use caution when inputting case-specific facts into any large language model. Generative AI apps, such as ChatGPT, allow developers to see and potentially use information you input to further train the app. This disclosure could constitute a breach of your duty of confidentiality and risk your client's confidential information becoming public. • Ensure client needs are met. ChatGPT and other generative large language models can be a helpful tool litigators can leverage in their daily case work. Note, though, these models are not reading and interpreting cases or secondary sources to provide an informed response to your request. Instead, the apps use machine learning to predict the most likely next word in a sentence. Accordingly, be sure to supplement all generative AI outputs with your own research and analysis to meet your client's legal needs. • Understand plagiarism risks. Litigators may be tempted to use generative AI apps to drafts briefs, pleadings, and other memoranda. However, any large language model's sources are not readily apparent to end users. When drafting legal documents, use generative AI tools with caution to ensure you are not plagiarizing an existing source and exposing you or your firm to liability. For additional resources on generative AI, see Generative Artificial Intelligence (AI) Resource Kit. For a full listing of key content covering fundamental civil litigation tasks throughout a federal court litigation lifecycle, see Civil Litigation Fundamentals Resource Kit (Federal).