Methodology
Oklahoma Rule 1.17 AI Compliance Guide
Verification framework for AI-assisted commutation documents
Abstract
This document describes Commutation.app’s methodology for ensuring that AI-assisted commutation application materials comply with Oklahoma Court of Criminal Appeals Rule 1.17 (Use of Generative AI). Rule 1.17, adopted February 18, 2026, requires that any portion of a document produced or modified by generative AI must be verified as accurate by the person responsible for the document before filing with the OCCA.
While commutation applications are filed with the Oklahoma Pardon and Parole Board (PPB) rather than the OCCA, Commutation.app adopts Rule 1.17’s verification standards as a best-practice framework for all AI-assisted legal document preparation. This methodology describes the five-phase verification process, the technical controls that enforce it, and the audit trail generated for compliance documentation.
Background
The use of AI tools in legal document preparation raises significant ethical and professional responsibility questions. Oklahoma CCA Rule 1.17 is the first AI-specific rule adopted by any Oklahoma court, establishing a clear standard: human verification of all AI-generated content before filing.
Commutation applications occupy a unique position in the legal landscape. They are administrative proceedings before the Pardon and Parole Board, not court filings in the traditional sense. Rule 1.17 technically applies only to OCCA filings. However, the principles underlying Rule 1.17—human verification, accuracy assurance, and responsible AI use—are equally important in commutation applications, where the stakes involve an individual’s liberty.
Commutation.app uses AI to assist with document organization, information retrieval, and application structuring. The AI Copilot helps users draft narrative sections, organize supporting evidence, and structure their applications for PPB review. Because these AI-assisted outputs may ultimately be presented to the PPB and could affect liberty interests, Commutation.app voluntarily applies Rule 1.17’s verification standard to all AI-generated content on the platform.
Methodology
Phase 1: AI Content Identification
The first phase identifies which portions of the commutation application were produced or modified by AI. Commutation.app tracks AI involvement at the field level:
- User-entered content: Information typed directly by the user (personal details, DOC number, case information) is marked as human-authored.
- Auto-filled content: Data populated from ODOC or OSCN records is marked as system-sourced with the data origin noted.
- AI-assisted content: Narrative sections drafted or refined by the AI Copilot are marked as AI-generated with the specific model and timestamp recorded.
- AI-modified content: Sections where the user drafted text and the AI suggested modifications are marked as AI-modified, preserving both the original and modified versions.
Phase 2: Factual Verification
The second phase verifies the accuracy of factual assertions in AI-generated content. This includes:
- ODOC record cross-reference: Sentence information, facility assignment, and institutional records cited in the application are verified against publicly available ODOC data.
- OSCN case verification: Court case numbers, conviction details, and sentencing information are verified against Oklahoma State Courts Network records.
- Statistical claims: Any statistics cited by the AI (incarceration rates, commutation grant rates, recidivism data) are verified against their cited sources and flagged if the source cannot be confirmed.
- Date verification: All dates referenced in the application (arrest dates, conviction dates, program completion dates) are cross-referenced against the case record.
Phase 3: Legal Accuracy Review
The third phase evaluates the legal accuracy of AI-generated content:
- Eligibility assessment accuracy: If the AI Copilot generated eligibility analysis, the legal basis for the eligibility determination is verified against current Oklahoma statutes and PPB rules.
- Statutory citations: Any Oklahoma statutes referenced in the application are verified for current text and applicability.
- Procedural accuracy: Statements about the PPB process, hearing procedures, or submission requirements are verified against current PPB rules and procedures.
Phase 4: Human Review Gate
The fourth phase is a mandatory human review step. No AI-generated content can be included in the final application PDF without explicit human approval:
- Section-by-section review: The application preview highlights all AI-generated or AI-modified sections, requiring the user to review and explicitly approve each section.
- Verification prompts: For each AI-generated section, the user is prompted to confirm that the information is accurate and complete to the best of their knowledge.
- Edit capability: Users can edit any AI-generated content before approval. Edits are tracked and the modified content is re-classified as human-authored.
Phase 5: Audit Trail Generation
The fifth phase generates a compliance audit trail documenting the verification process:
- AI content inventory: A record of which sections were AI-generated, which model was used, and when the content was generated.
- Verification log: Timestamps and results for each verification step (ODOC cross-reference, OSCN verification, statutory checks).
- Human approval record: Timestamps and user identity for each section approval.
- Edit history: A diff log of all changes made to AI-generated content during the review process.
The audit trail is stored securely and associated with the application record. It can be produced if compliance is ever questioned by the PPB, an attorney, or any reviewing body.
Data Sources
- Oklahoma CCA Rule 1.17: The primary regulatory authority establishing AI verification requirements for OCCA filings. Adopted February 18, 2026, by the Oklahoma Court of Criminal Appeals.
- Oklahoma Rules of Professional Conduct: Rule 1.1 (Competence), Rule 1.6 (Confidentiality), Rule 3.3 (Candor Toward the Tribunal), and Rule 5.5 (Unauthorized Practice of Law) inform the ethical framework applied to AI-assisted legal document preparation.
- ODOC Public Records: Oklahoma Department of Corrections public records used for factual verification of incarceration data, sentence information, and institutional records.
- OSCN (Oklahoma State Courts Network): Court records used for case verification, conviction details, and sentencing information cross-referencing.
- PPB Rules and Procedures: Current Oklahoma Pardon and Parole Board procedural requirements for commutation applications.
Limitations and Disclaimers
- Voluntary adoption: Oklahoma CCA Rule 1.17 applies to OCCA filings, not PPB commutation applications. Commutation.app’s adoption of Rule 1.17 standards is voluntary and represents a best-practice approach, not a legal requirement for commutation applications.
- Verification scope: Automated verification checks are limited to publicly available data sources (ODOC public records, OSCN). Information that is not publicly available (e.g., sealed records, non-digitized documents) cannot be verified through automated means.
- Human responsibility: The compliance framework assists with verification but does not replace human responsibility. The person filing the commutation application is ultimately responsible for the accuracy of all content, regardless of whether AI was used in its preparation.
- Not legal advice: Commutation.app is not a law firm and does not provide legal advice. This compliance methodology does not constitute a legal opinion on Rule 1.17 compliance. Consult a licensed attorney for advice specific to your situation. See our UPL Disclaimer for important limitations.
Last updated: March 2026
See also: Parole Eligibility Assessment Methodology | Readiness Checklist