Oklahoma AI Ethics Rules for Lawyers

The official instruments

A rule from the criminal side of the split bench

Oklahoma runs two courts of last resort, the Supreme Court for civil matters and the Court of Criminal Appeals for criminal ones, and it is the criminal court that wrote the state’s binding AI text. By order published as 2026 OK CR 7 and effective on its date, February 18, 2026, the OCCA adopted Rule 1.17, Use of Generative Artificial Intelligence, into its rules. The tracker records it as a court rule and notes the milestone precisely: this is the first criminal-appellate AI rule recorded here, a category no other jurisdiction on the states index yet occupies. Scope is the fact to hold onto before anything else. Rule 1.17 governs filings before the OCCA, which means it binds criminal appellate practice statewide while leaving civil litigation and the district courts to the baseline duties, an inversion of the usual pattern where civil-side rulemaking arrives first.

Verified as accurate, by a person, with a name on the line

The rule’s mechanism is a human checkpoint. When generative AI is used in drafting any OCCA filing, the party or counsel must ensure that every AI-produced or AI-modified portion has been verified as accurate by a person responsible for the document. Note what the drafting choice accomplishes: verification is not an abstract quality standard but an act someone identifiable performs, which forecloses the everyone-assumed-someone-checked defense that runs through the sanctions caselaw. Noncompliance carries a menu ending in contempt, and starting somewhere more expensive for the client: waiver of the affected issues, then striking of the document. In criminal appeals, waiver means an argument for a person’s liberty can die because nobody verified the passage advancing it. Presiding Judge Lumpkin specially concurred, making the point that a signature on a filing already certifies its accuracy, a reminder that Rule 1.17 makes explicit what signing always implied.

The bar’s resources are real; its opinion shelf is empty

The Oklahoma Bar Association has issued no AI ethics opinion. Its ethics infrastructure exists, an ethics counsel and the LEAP advisory panel that fields members’ questions, but nothing AI-specific had issued from either at our July 1, 2026 check, so the OCCA rule stands alone in Oklahoma’s instrument column. The bar’s Management Assistance Program has published articles on AI by bar staff; they are undated on the page, and they are education, so this tracker records them as activity rather than guidance, the same honest line drawn for bar-magazine material everywhere. One more item sits in the pipeline: an Oklahoma Supreme Court decision published as 2026 OK 19 has surfaced in our review queue, and because its case name is unresolved on our side, this page records only that it exists and characterizes nothing about it until the primary is read, per the methodology.

Everything Rule 1.17 does not cover

For the Oklahoma lawyer outside an OCCA docket, the governing law is the familiar baseline. Confidentiality analysis before client information reaches a tool, per the Rule 1.6 breakdown. Verification before anything is filed anywhere, because Rule 1.17’s logic is a preview of how any court reacts to fabricated authority. Honest billing per the fee ethics analysis, supervision per the supervision rules, and a written office policy per the policy template, with ABA Formal Opinion 512 supplying the national frame. One federal note, clearly labeled as federal and outside this state dataset: AI-related requirements have also surfaced in Oklahoma’s federal district courts, the Eastern and Western Districts among them, which run on their own orders, the venue-by-venue check the court orders guide covers. The changelog will record the day Oklahoma’s civil side or its bar produces an instrument of its own.