Federal Courts Split on Generative AI in Legal Filings
Federal courts are adopting three distinct approaches to generative AI in litigation, creating a patchwork of rules that attorneys must navigate based on jurisdiction. Some courts require detailed disclosure and certification of AI use. Others ban it outright. A third group relies on existing professional responsibility rules without new procedures.
The variation reflects deeper disagreement about whether AI requires new judicial infrastructure or whether traditional attorney accountability standards suffice.
Three Regulatory Models
Mandatory Disclosure and Certification: The most common approach requires attorneys to disclose AI use, identify the specific tool, flag which sections are AI-generated, and certify that a human attorney reviewed the content and verified citations. The U.S. District Court for the Middle District of Pennsylvania follows this model.
Outright Prohibition: The Northern District of Ohio bans generative AI for preparing court filings entirely. The court permits legal research tools like Westlaw and LexisNexis, as well as public search engines, but not AI drafting assistants.
Responsibility-Based Standard: Courts including the Southern District of California treat AI-assisted filings under Federal Rule of Civil Procedure 11, which requires attorneys to review and verify all content. No separate AI declaration is required.
Eastern District of Missouri: The Responsibility-Based Model in Practice
The Eastern District of Missouri exemplifies the responsibility-based approach. Its general policy holds filers accountable for all content, including AI-generated portions, under Rule 11(b). The signature on a filing signifies that the attorney has verified its accuracy and legitimacy.
The court does not require pre-filing certification, identification of AI tools, flagging of AI-drafted sections, or AI-specific sanctions. Sanctions arise solely from the general Rule 11 framework.
One exception exists: Judge Joshua M. Divine, who sits in both the Eastern and Western Districts of Missouri, requires all litigants to file a certificate indicating either that no generative AI was used or that any AI-generated language was checked for accuracy.
How Eastern District of Missouri Differs From Peers
Disclosure: Unlike most courts with AI protocols, Eastern District of Missouri does not require affirmative disclosure. The attorney's signature is sufficient.
Attorney duties: Courts with mandatory protocols require detailed verification and record-keeping. Eastern District of Missouri relies on longstanding Rule 11(b) standards, requiring reasonable inquiry into facts and law supporting the filing.
Sanctions: Other courts tie sanctions to AI-specific violations. Eastern District of Missouri's sanctions arise from general Rule 11 violations, making no distinction based on AI use.
Alignment: The District of Connecticut and Southern District of Texas follow the same approach, relying on established professional responsibility rules.
Practical Implications for Multi-Jurisdictional Practice
Attorneys managing cases across multiple districts face compliance challenges. A filing acceptable in Eastern District of Missouri may violate requirements in the Middle District of Pennsylvania.
The safest practice is to follow the most stringent applicable standard: rigorously verify AI outputs, document the review process, and disclose AI use even if not explicitly required by local rule. All courts demand accurate, well-researched filings regardless of their AI policies.
For attorneys seeking to understand how AI fits into legal practice, resources on AI for Legal professionals cover document review, legal research automation, and compliance requirements across different jurisdictions.
AI is a tool. Responsibility for its output remains with the attorney.
Your membership also unlocks:
