Florida's New AI Certification Rule (SC2026-0673): What Every Attorney Must Certify Now

On May 28, 2026 the Florida Supreme Court amended statewide rules to require attorneys to certify that cited authorities are accurate. Here is exactly what you now must certify, and how AI-fabricated citations trigger sanctions.

What Florida's New AI Rule Actually Requires

On May 28, 2026, on its own motion, the Florida Supreme Court adopted the amendment in case number SC2026-0673, which modifies Rule of General Practice and Judicial Administration 2.515(d)(2). The amendment is effective June 15, 2026, and applies to both attorneys and self-represented (pro se) litigants. It directly addresses the use of artificial intelligence in preparing court filings and the problem of AI-generated, non-existent legal citations, often called "hallucinations."

The core of the amendment is a new certification. By signing a document filed with the court, the signer now represents that "the legal authorities identified exist and are accurately cited." The rule expressly authorizes courts to impose appropriate sanctions for any filing inconsistent with that representation, which includes briefs containing fabricated or inaccurately cited authorities, such as the convincing but nonexistent cases that generative AI tools can produce. The Court adopted the rule principally to create a single statewide standard in place of the patchwork of circuit administrative orders (Miami-Dade's AO 26-04, Palm Beach County, and others) that had imposed varied AI disclosure and certification requirements. The order makes clear that ignorance is not a defense. The signing attorney is strictly responsible for the validity of every citation submitted.

Rule 4-1.1 Competence and the AI Verification Mandate

The new certification requirement is a direct extension of an attorney's existing duty of competence under Rule 4-1.1 of the Rules Regulating The Florida Bar. Competent representation requires the legal knowledge, skill, thoroughness, and preparation reasonably necessary for the representation. The comment to ABA Model Rule 1.1, upon which Florida's rule is based, was updated years ago to include technological competence. It specifies that lawyers must keep abreast of the benefits and risks associated with relevant technology.

Submitting a filing with fabricated case law is a fundamental failure of competence. The new amendment makes this explicit. It is no longer enough to simply trust the output of a generative AI tool like ChatGPT, CoCounsel, or Harvey. An attorney must now have a verifiable process for checking every single citation.

For a 12-attorney personal injury firm in Tampa, this changes the workflow for drafting a motion for summary judgment. Previously, a paralegal might have used a cloud-based AI tool to generate a first draft, summarizing relevant case law on proximate cause. The partner would then review the draft for legal argument. Now, that partner must also personally verify that *Smith v. Jones*, 123 So. 3d 456, as cited by the AI, is a real case, that the citation is correct, and that the case actually supports the proposition for which it is cited. This requires direct access to a traditional legal research platform like Westlaw or LexisNexis, or a reliable internal system, for every citation produced by an AI assistant.

Rule 4-5.3 Supervising Nonlawyer Assistants (Including AI)

Florida Rule 4-5.3 requires lawyers to make reasonable efforts to ensure that the conduct of nonlawyer assistants is compatible with the professional obligations of the lawyer. The official comments clarify that this includes assistants like secretaries, investigators, and paraprofessionals. The Florida Bar has been clear that when an AI tool is used to perform substantive legal work, it functions as a nonlawyer assistant.

Therefore, the duty of supervision applies directly to the AI's work product. A managing attorney cannot delegate the task of AI-driven research to a junior associate or paralegal without also supervising their verification process. The ultimate responsibility for the accuracy of the final filing remains with the signing attorney, as reinforced by the new certification under Rule 2.515(d)(2).

Consider an immigration practice with three offices in South Florida handling complex asylum cases. They might use an AI tool to research country conditions and cite to international treaties or foreign court decisions. Under Rule 4-5.3 and the new certification mandate, the supervising partner must ensure the paralegal using the AI has a protocol to cross-reference every generated citation against a primary source database. Simply accepting the AI's output and incorporating it into a brief for the Executive Office for Immigration Review is a direct violation and now exposes the attorney to sanctions. The State Bar of California's recent guidance on generative AI, issued in late 2023, similarly stressed this supervisory duty, highlighting the risk of inadvertently disclosing confidential client information and the need to verify AI outputs for accuracy.

Rule 4-1.6 Confidentiality and the Cloud AI Problem

While the new certification rule focuses on citation accuracy, the underlying choice of AI tools implicates a more profound ethical duty: confidentiality under Rule 4-1.6. Most widely publicized legal AI tools, including CoCounsel, Harvey, and Spellbook, are cloud-based. This means that when a firm uploads a client's medical records, a draft settlement agreement, or deposition transcripts for analysis, that confidential data is transmitted to and processed on third-party servers.

This act of transmission may breach the duty of confidentiality if not handled with extreme care and explicit client consent. ABA Formal Opinion 477R requires attorneys to exercise reasonable efforts to prevent the inadvertent or unauthorized disclosure of client information. Sending sensitive case data to a vendor whose data handling, security protocols, and subsequent use of that data are not fully understood is a significant risk. The New York State Bar Association's AI ethics guidance from early 2024 specifically warned attorneys about the confidentiality risks of using public generative AI tools, which may use client data to train their models.

For a Florida family law practice, uploading a client's financial affidavit or a child custody evaluation into a cloud AI tool is fraught with peril. It potentially exposes highly sensitive, personally identifiable information and privileged communications to a third-party's infrastructure. This is a risk many firms, particularly those with a HIPAA-adjacent compliance posture, have correctly identified as unacceptable.

AI Filing Certification: A Compliance Checklist

Rule / Requirement	What It Requires	Compliant Action
Rule 2.515(d)(2) Certification	Attorney signature certifies all legal citations have been verified for accuracy and are not fictitious.	Manually check every AI-generated citation in a primary legal database (Westlaw, Lexis) before filing.
Rule 4-1.1 Competence	Maintain the requisite legal knowledge and skill, including understanding the risks of technology.	Understand how your firm's AI tool works. Do not "trust" its output without independent verification.
Rule 4-5.3 Supervision	Supervise the work of nonlawyer assistants, including AI tools, to ensure compliance with professional obligations.	Implement and enforce a firm-wide policy requiring human verification of all AI-generated legal research and citations.
Rule 4-1.6 Confidentiality	Prevent the inadvertent or unauthorized disclosure of information relating to the representation of a client.	Avoid uploading confidential client documents to third-party cloud AI services. Use on-premise solutions where possible.

What Mi Assist Legal Does

Our team at Mi Assist Legal provides OpenClaw Legal, an on-premise AI document search and analysis tool. It installs locally on a Mac Mini or Docker host inside your firm's network, ensuring no client data ever leaves your control. By running models like Llama 3.2 locally via Ollama and indexing your case files into a private ChromaDB or Qdrant database, we eliminate the confidentiality risks of cloud-based AI that trigger Rule 4-1.6 concerns. Every answer OpenClaw provides is cited directly to the source documents in your own files, giving you a clear, verifiable path to compliance with Florida's new certification rule.

Frequently Asked Questions

What are the potential sanctions for violating the new certification rule?

Sanctions can range from the court striking the filing to monetary penalties, referral to The Florida Bar for disciplinary action, and potential findings of bad faith conduct, depending on the severity and intent of the violation.

Does this rule mean we cannot use AI at all?

No. The rule does not prohibit the use of AI. It mandates accountability. You can use AI for drafting, research, and summarization, but you, the attorney, are required to independently verify the accuracy of all legal citations before filing.

Is my paralegal allowed to use AI to help draft a motion?

Yes, but under your direct supervision per Rule 4-5.3. You must have a process in place to review and verify their work product, including every case citation generated by the AI tool they used. The ultimate responsibility for the filing's accuracy rests with you.

How is an on-premise AI tool different from a cloud one like CoCounsel?

An on-premise tool like our OpenClaw Legal system runs entirely on a server within your physical office. Your firm's data is never sent over the internet to a third party. Cloud tools require you to upload your data to the vendor's servers, creating confidentiality and data security risks under Rule 4-1.6.

To ensure your firm's AI adoption strategy is fully compliant with your ethical obligations under the new Florida rules, schedule a 30-minute on-site assessment with our team. We can demonstrate how a fully local, verifiable AI assistant provides the efficiency you need without compromising client confidentiality or professional responsibility.