SF4636
License establishment for artificial intelligence independent verification organizations
Legislative Session 94 (2025-2026)
Related bill: HF4544
Bill text
(underlined = new law, struck through = current law the bill removes)
Versions
1.1A bill for an act
1.2relating to commerce; establishing a license for artificial intelligence independent
1.3verification organizations; establishing an advisory council; authorizing rulemaking;
1.4requiring reports; proposing coding for new law in Minnesota Statutes, chapter
1.5325M.
1.6BE IT ENACTED BY THE LEGISLATURE OF THE STATE OF MINNESOTA:
1.7Section 1. new text begin [325M.50] DEFINITIONS. new text end
1.8 new text begin Subdivision 1. Scope. For purposes of sections 325M.51 to 325M.54, the following new text end
1.9 new text begin terms have the meanings given. new text end
1.10 new text begin Subd. 2. Artificial intelligence application. "Artificial intelligence application" means new text end
1.11 new text begin a software program or system that uses artificial intelligence to perform tasks that typically new text end
1.12 new text begin require human intelligence. new text end
1.13 new text begin Subd. 3. Artificial intelligence model. "Artificial intelligence model" means an new text end
1.14 new text begin engineered or machine-based system that can, for explicit or implicit objectives, infer from new text end
1.15 new text begin an input received how to generate output that can influence physical or virtual environments. new text end
1.16 new text begin Subd. 4. Commissioner. "Commissioner" means the commissioner of commerce. new text end
1.17 new text begin Subd. 5. Deployer. "Deployer" means a person that implements, integrates, or makes new text end
1.18 new text begin operational an artificial intelligence model or artificial intelligence application in this state. new text end
1.19 new text begin Deployer includes a person that makes an artificial intelligence model or artificial intelligence new text end
1.20 new text begin application available for use by others in this state, whether directly or as part of a product new text end
1.21 new text begin or service. new text end
2.1 new text begin Subd. 6. Developer. "Developer" means a person that develops an artificial intelligence new text end
2.2 new text begin model or artificial intelligence application that is deployed in this state. new text end
2.3 new text begin Subd. 7. Independent verification organization or IVO. "Independent verification new text end
2.4 new text begin organization" or "IVO" means an entity licensed by the commissioner pursuant to section new text end
2.5 new text begin 325M.51 to assess an artificial intelligence model's or artificial intelligence application's new text end
2.6 new text begin adherence to standards reflecting best practices for the prevention of personal injury and new text end
2.7 new text begin property damage. new text end
2.8 new text begin Subd. 8. Security vendor. "Security vendor" means a third-party entity engaged by an new text end
2.9 new text begin IVO or developer to evaluate the safety or security of an artificial intelligence model or new text end
2.10 new text begin artificial intelligence application, including by using processes such as risk detection and new text end
2.11 new text begin risk mitigation. new text end
2.12Sec. 2. new text begin [325M.51] INDEPENDENT VERIFICATION ORGANIZATION LICENSURE new text end
2.13 new text begin REQUIREMENT. new text end
2.14 new text begin Subdivision 1. License required. An application for an independent verification new text end
2.15 new text begin organization license must be made by filing with the commissioner the information, materials, new text end
2.16 new text begin and forms specified in rules adopted by the commissioner. new text end
2.17 new text begin Subd. 2. IVO proposed plan. An IVO must file a proposed plan with the IVO's new text end
2.18 new text begin application. The proposed plan must include: new text end
2.19 new text begin (1) the risk or risks that artificial intelligence models or artificial intelligence applications new text end
2.20 new text begin must mitigate. For each risk identified, the plan must include: new text end
2.21 new text begin (i) a proposed definition of acceptable levels of risk; new text end
2.22 new text begin (ii) metrics that are measurable and can be used to determine whether the acceptable new text end
2.23 new text begin level of risk defined by the IVO produces beneficial outcomes; new text end
2.24 new text begin (iii) target levels for the metrics, including data sources the target levels are based on new text end
2.25 new text begin and methods for measurement; and new text end
2.26 new text begin (iv) a description of the evaluation and reporting protocol to determine whether verified new text end
2.27 new text begin artificial intelligence models or artificial intelligence applications meet the outcome metrics new text end
2.28 new text begin on an ongoing basis; new text end
2.29 new text begin (2) proposed technical, operational, governance, and other mitigation requirements for new text end
2.30 new text begin developers or deployers, including procedures for predevelopment and postdevelopment, new text end
2.31 new text begin to ensure acceptable levels of risk, including: new text end
2.32 new text begin (i) ongoing monitoring risks; and new text end
3.1 new text begin (ii) ongoing assessment of mitigation efficacy; new text end
3.2 new text begin (3) methodologies and sources used to evaluate the efficacy of mitigation requirements; new text end
3.3 new text begin (4) benchmarks, technologies, and audit methodologies proposed to assess developer new text end
3.4 new text begin and deployer adherence to mitigation requirements; new text end
3.5 new text begin (5) an approach to assessing continued good standing of a verified artificial intelligence new text end
3.6 new text begin model or artificial intelligence application, including reviewing and evaluating the developer's new text end
3.7 new text begin or deployer's maintenance of artificial intelligence governance plans and policies, processes new text end
3.8 new text begin for risk monitoring and mitigation, whistleblower protections, and training for employees new text end
3.9 new text begin and third parties; new text end
3.10 new text begin (6) disclosure requirements for developers or deployers related to detected risks, incident new text end
3.11 new text begin reports, or material changes to risk profiles, including both risks detected before verification new text end
3.12 new text begin and risks resulting from fine-tuning or modifying an artificial intelligence model or artificial new text end
3.13 new text begin intelligence application after verification; new text end
3.14 new text begin (7) procedures for prescribing and verifying implementation of corrective actions to new text end
3.15 new text begin remedy a developer's or deployer's identified failure to: new text end
3.16 new text begin (i) achieve an acceptable level of risk with respect to an artificial intelligence application new text end
3.17 new text begin or artificial intelligence model; new text end
3.18 new text begin (ii) comply with any other mitigation requirements promulgated by the applicant; and new text end
3.19 new text begin (iii) comply with the developer's or deployer's artificial intelligence governance plan new text end
3.20 new text begin and policy; new text end
3.21 new text begin (8) standards and procedures for revoking verification for noncompliance with the new text end
3.22 new text begin applicant's mitigation requirements, failure to achieve acceptable levels of risk, or new text end
3.23 new text begin noncompliance with the developer's or deployer's artificial intelligence governance plans new text end
3.24 new text begin and policies; new text end
3.25 new text begin (9) whether the applicant proposes risk-specific verification and how plans are tailored new text end
3.26 new text begin to the specific risk; new text end
3.27 new text begin (10) coordination with federal and state authorities; new text end
3.28 new text begin (11) personnel qualifications; new text end
3.29 new text begin (12) governance policies, sources of funding, and policies ensuring independence; and new text end
3.30 new text begin (13) any other information required by the commissioner. new text end
4.1 new text begin Subd. 3. License determination. (a) The commissioner may license an applicant if the new text end
4.2 new text begin commissioner finds: new text end
4.3 new text begin (1) the applicant demonstrated independence from the artificial intelligence community; new text end
4.4 new text begin and new text end
4.5 new text begin (2) every element of the applicant's proposed plan is adequate to ensure that artificial new text end
4.6 new text begin intelligence models or artificial intelligence applications verified pursuant to the plan mitigate new text end
4.7 new text begin one or more risks to an acceptable level. new text end
4.8 new text begin (b) If verification is proposed to a specific risk, the commissioner must evaluate the new text end
4.9 new text begin plan's adequacy accordingly. new text end
4.10 new text begin (c) If the commissioner finds that an applicant's plan does not adequately mitigate all of new text end
4.11 new text begin the proposed risks, the applicant must be licensed to verify only those risks for which the new text end
4.12 new text begin plan is deemed adequate. new text end
4.13 new text begin (d) The license must specify: new text end
4.14 new text begin (1) the risks the IVO is authorized to verify; and new text end
4.15 new text begin (2) any markets the license applies to. new text end
4.16 new text begin Subd. 4. License revocation. The commissioner must revoke an IVO license if the new text end
4.17 new text begin commissioner determines: new text end
4.18 new text begin (1) the IVO's plan is materially misleading or inaccurate; new text end
4.19 new text begin (2) the IVO fails to adhere to the IVO's plan in a way that materially impairs the IVO's new text end
4.20 new text begin responsibilities, including failure to adhere to the plan's procedures for ongoing monitoring new text end
4.21 new text begin of verified artificial intelligence models or artificial intelligence applications and new text end
4.22 new text begin implementation of corrective action; new text end
4.23 new text begin (3) a material change compromises independence from the artificial intelligence industry; new text end
4.24 new text begin (4) technological evolution renders methods obsolete for ensuring acceptable levels of new text end
4.25 new text begin the risk the commissioner has designated the IVO to verify; or new text end
4.26 new text begin (5) a verified artificial intelligence model or artificial intelligence application causes new text end
4.27 new text begin material harm of the type the IVO defines an acceptable level of risk in order to prevent. new text end
4.28 new text begin Subd. 5. Cure opportunity. The commissioner may allow an IVO to cure the basis for new text end
4.29 new text begin revocation before terminating the license. new text end
4.30 new text begin Subd. 6. Fees. The commissioner must establish reasonable application and renewal new text end
4.31 new text begin fees sufficient to offset administrative costs. new text end
5.1 new text begin Subd. 7. Verification not required. Nothing in this section requires an artificial new text end
5.2 new text begin intelligence model or artificial intelligence application to seek IVO verification. new text end
5.3 new text begin Subd. 8. Rulemaking. The commissioner may adopt rules necessary to implement new text end
5.4 new text begin sections 325M.51 to 325M.54. new text end
5.5Sec. 3. new text begin [325M.52] INDEPENDENT VERIFICATION ORGANIZATION new text end
5.6 new text begin REQUIREMENTS. new text end
5.7 new text begin Subdivision 1. Implementation. A licensed IVO must implement the IVO's approved new text end
5.8 new text begin plan. new text end
5.9 new text begin Subd. 2. Revocation of verification. An IVO must revoke verification if a developer new text end
5.10 new text begin or deployer: new text end
5.11 new text begin (1) fails to meet mitigation requirements; new text end
5.12 new text begin (2) fails to cooperate with monitoring; new text end
5.13 new text begin (3) violates governance policies; or new text end
5.14 new text begin (4) fails to implement corrective actions. new text end
5.15 new text begin Subd. 3. Plan modification. (a) An IVO may update or modify: new text end
5.16 new text begin (1) technical and operational requirements; new text end
5.17 new text begin (2) evaluation benchmarks; new text end
5.18 new text begin (3) audit methodologies; new text end
5.19 new text begin (4) governance plans; new text end
5.20 new text begin (5) verification activities to enhance efficacy; and new text end
5.21 new text begin (6) any other element of the IVO's plan in order to take advantage of improved new text end
5.22 new text begin technology. new text end
5.23 new text begin (b) An IVO may address previously discovered issues with the IVO's plan. new text end
5.24 new text begin (c) A notice of material changes must be reported in writing to the commissioner. The new text end
5.25 new text begin notice shall describe the changes, the rationale for the changes, and an explanation of how new text end
5.26 new text begin the changes better enable the IVO to ensure the requisite level of mitigation of relevant new text end
5.27 new text begin risks. new text end
5.28 new text begin (d) Changes must take effect upon notification. new text end
6.1 new text begin (e) The commissioner may, within six months after receiving notice of changes, request new text end
6.2 new text begin additional information from the IVO regarding the changes or may issue a written notice new text end
6.3 new text begin rejecting the changes in whole or in part. If the commissioner rejects the changes, the IVO new text end
6.4 new text begin has 30 days to modify the IVO's plan to comply with the commissioner's determination and new text end
6.5 new text begin to assess whether artificial intelligence models or applications assessed under the previous new text end
6.6 new text begin plan must be reassessed. new text end
6.7 new text begin Subd. 4. Annual reporting. (a) An IVO must submit an annual report to the new text end
6.8 new text begin commissioner and chairs and ranking minority members of the legislative committees with new text end
6.9 new text begin jurisdiction over artificial intelligence. The report must include: new text end
6.10 new text begin (1) aggregated information on the capabilities of the artificial intelligence models and new text end
6.11 new text begin artificial intelligence applications evaluated by the IVO, the observed societal risks and new text end
6.12 new text begin benefits associated with those capabilities, and the potential societal risks and benefits new text end
6.13 new text begin associated with those capabilities; new text end
6.14 new text begin (2) the adequacy of evaluation resources, technical capabilities, and mitigation measures new text end
6.15 new text begin to address observed and potential risks; new text end
6.16 new text begin (3) aggregated results of verification assessments; new text end
6.17 new text begin (4) aggregated and anonymized compliance with prescribed remediation; new text end
6.18 new text begin (5) anonymized descriptions of any additional significant risk the IVO observed while new text end
6.19 new text begin conducting assessments, even if the risk is not one the IVO is licensed to verify; new text end
6.20 new text begin (6) a list of verified artificial intelligence models and artificial intelligence applications; new text end
6.21 new text begin (7) a description of evaluation methods; and new text end
6.22 new text begin (8) governance or funding changes that affect independence. new text end
6.23 new text begin (b) IVOs may redact trade secrets, sensitive business information, personally identifiable new text end
6.24 new text begin information, and other security-sensitive content. new text end
6.25 new text begin (c) Documentation used in reports must be retained for ten years. An IVO must also new text end
6.26 new text begin retain all documentation relating to the IVO's assessment and verification of artificial new text end
6.27 new text begin intelligence models or artificial intelligence applications, including ongoing monitoring and new text end
6.28 new text begin any subsequent corrective action, for ten years after the relevant activity. new text end
6.29 new text begin (d) The commissioner must publish on the Department of Commerce website redacted new text end
6.30 new text begin versions of reports issued by IVOs. new text end
7.1Sec. 4. new text begin [325M.53] ARTIFICIAL INTELLIGENCE ADVISORY COUNCIL. new text end
7.2 new text begin Subdivision 1. Establishment. (a) The council is established within the Department of new text end
7.3 new text begin Commerce. The commissioner must determine the appropriate size of the council and appoint new text end
7.4 new text begin all members. new text end
7.5 new text begin (b) Membership must include at least one civil society representative, including but not new text end
7.6 new text begin limited to nongovernmental organizations, educational and research institutions, public new text end
7.7 new text begin policy institutes, or consumer and business advocacy organizations. new text end
7.8 new text begin Subd. 2. Duties. (a) The commissioner must delegate powers, including licensing and new text end
7.9 new text begin auditing, to the council. new text end
7.10 new text begin (b) A member must: new text end
7.11 new text begin (1) remain free of undue influence and refrain from any action that could compromise new text end
7.12 new text begin the member's ability to carry out the member's responsibilities or otherwise cast doubt on new text end
7.13 new text begin the member's ability to independently assess artificial intelligence models or artificial new text end
7.14 new text begin intelligence applications; new text end
7.15 new text begin (2) refrain from any action or occupation, gainful or not, that is incompatible with the new text end
7.16 new text begin member's duties, including but not limited to employment by a developer or deployer of new text end
7.17 new text begin artificial intelligence; new text end
7.18 new text begin (3) refrain from owning or acquiring any equity or other interest, directly or indirectly, new text end
7.19 new text begin in a company whose business consists in significant part of developing or deploying artificial new text end
7.20 new text begin intelligence; new text end
7.21 new text begin (4) observe a one-year postemployment restriction from artificial intelligence firms or new text end
7.22 new text begin IVOs; and new text end
7.23 new text begin (5) be qualified to assess IVO plans. new text end
7.24 new text begin Subd. 3. Administration. (a) A member must not serve more than two consecutive new text end
7.25 new text begin terms. new text end
7.26 new text begin (b) A member must receive reimbursement for actual and necessary expenses incurred new text end
7.27 new text begin in the discharge of duties. A member may also receive a salary for carrying out the member's new text end
7.28 new text begin duties under this section. new text end
7.29 new text begin (c) A member may be removed for inefficiency, neglect, or malfeasance. new text end
7.30 new text begin (d) A majority of members constitutes a quorum and concurrence of a majority of a new text end
7.31 new text begin quorum is sufficient for the council's determination. new text end
8.1 new text begin (e) The council must keep a record of the council's proceedings, including all new text end
8.2 new text begin considerations relating to the issuance, refusal, renewal, and revocation of an IVO license. new text end
8.3Sec. 5. new text begin [325M.54] LIMITED LIABILITY; REBUTTABLE PRESUMPTION. new text end
8.4 new text begin In a civil action asserting claims for personal injury or property damage caused by an new text end
8.5 new text begin artificial intelligence model or artificial intelligence application, there is a rebuttable new text end
8.6 new text begin presumption against liability if: new text end
8.7 new text begin (1) the artificial intelligence model or artificial intelligence application in question was new text end
8.8 new text begin verified by a licensed IVO at the time of the plaintiff's alleged injury; new text end
8.9 new text begin (2) the plaintiff's alleged injury arose from a risk that the IVO was licensed to verify new text end
8.10 new text begin and for which the IVO did verify the artificial intelligence model or artificial intelligence new text end
8.11 new text begin application; and new text end
8.12 new text begin (3) the artificial intelligence model or artificial intelligence application is within the new text end
8.13 new text begin specified market segment, if any, for which the IVO was licensed to conduct verification. new text end
Bill information comes from public records published by the Office of the Revisor of Statutes, the Minnesota House and the Minnesota Senate. AI summaries are written by MinneSlate and are not official. MinneSlate is independent and is not affiliated with the State of Minnesota or the Minnesota Legislature.
Progress through the legislative process
In Committee