MIAMI-DADE, BROWARD COURTS ISSUE UNIFIED AI DISCLOSURE RULES

Stressing the need for uniformity, Miami-Dade and Broward courts on Wednesday jointly issued new administrative orders regarding the disclosure and verification of AI-generated court submissions.

“The Eleventh Judicial Circuit recognizes the importance of consistency and uniformity in the administration of justice concerning the disclosure of generative AI use in court filings and submissions,” 11th Judicial Circuit Chief Judge Ariana Fajardo Orshan said in a statement, adding that she was jointly issuing the order with the 17th Judicial Circuit.

In a similarly worded statement, 17th Judicial Circuit Chief Judge Carol-Lisa Phillips said she also recognized the importance of consistency and uniformity in AI policy.

“Our Administrative Order states that all citations, factual claims and conclusions must be verified and that those using AI must guard against potential plagiarism, confidentiality breaches, inaccurate information false information, hallucinations and the use of inappropriate language,” she said.

The 11th and 17th Circuits encompass Miami-Dade and Broward counties, respectively.

Lawyers and self-represented litigants who fail to comply with either of the orders face potential sanctions that range from the striking of the filing and denial of the requested relief to monetary sanctions and disciplinary referral.

In the 11th Circuit, Administrative Order No. 26-15, “Re: Use of Artificial Intelligence in Court Filings by Attorneys and Self-Represented Litigants; Disclosure of Use of Generative Artificial Intelligence,” rescinds previous Administrative Order No. 26-04.

The new order warns that AI tools may generate “inaccurate, misleading, or entirely fabricated factual assertions,” and reminds litigants of their duty of candor to the tribunal.

Quoting from a 2025 Third District Court of Appeal opinion, the order stresses that judges “should not have to parse case citations and parentheticals to discern whether cases exist.”

It goes on to provide an extensive definition of generative AI that concludes with a list of examples – “Harvey AI; Lexis+AI; AI.Law; Co-Counsel by Thomson Reuters; Westlaw drafting assistant; ChatGPT by OpenAI; Google Gemini; Microsoft Copilot; and Claude AI.”

Both orders reaffirm that the technology may be used for court submissions, but stress that “All information generated by AI must have appropriate human oversight that includes checking citations, verifying factual claims, and analyzing conclusions.”

Both orders require nearly all AI-generated court submissions to include a statement “substantially” in the following form:

• The undersigned hereby certifies that generative artificial intelligence was used in the preparation of this submission. The undersigned certifies that s/he has read and verified the accuracy of every citation to the law and/or the record, and the accuracy of any language drafted by the generative artificial intelligence, including quotations, citations, paraphrased assertions, facts, and legal analysis, and accepts full responsibility for the contents of this submission.

The orders list identical exceptions to disclosure.

They state that disclosure is not required when using AI “solely for routine research purposes, or for retrieving authorities or cite-checking,” or “when generative AI tools are used solely for grammar, spelling, or clarity edits, provided the substance and content of the filing is drafted and verified by the filer.”

In the 17th Judicial Circuit, “Administrative Order 2026-03-Gen (Amendment 2)” notes that generative AI “refers to a type of artificial intelligence that creates new, original content by learning and interpreting patterns from existing data.” The words “new, original content” are bold-faced.

“It operates by learning from large amounts of data to recognize and interpret patterns, enabling it to imitate human intelligence in tasks such as interpreting and producing language, generating citations, creating or analyzing images, and solving problems,” the definition states.

Subsection 5, “Confidentiality and Authenticity,” requires attorneys and self-represented litigants to “use their best efforts” to verify the authenticity of AI-generated information, “if there is a reasonable basis to question the authenticity and have produced inaccurate information, legal error, and/or AI (digital) hallucinations.”

The subsection continues, “AI tools for legal research, drafting documents, and assisting in the discovery process must be consistent with Florida law, court rules of procedure, the rules governing professional responsibilities, and the obligation to protect confidential information.”

Both orders contain identical provisions for sanctions.

“The failure to comply with this Administrative Order may result in sanctions, including but not limited to: the denial of the requested relief; contempt proceedings; striking of pleadings or dismissal of actions; monetary sanctions; referral to The Florida Bar or other appropriate authority; and any other sanction deemed appropriate by the Court, as permitted by law.”