HF4532
Artificial intelligence safety and disclosure requirements established, and civil remedies provided.
Legislative Session 94 (2025-2026)
Related bill: SF4509
Bill text
(underlined = new law, struck through = current law the bill removes)
Versions
1.1A bill for an act
1.2relating to commerce; establishing artificial intelligence safety and disclosure
1.3requirements; providing civil remedies; proposing coding for new law in Minnesota
1.4Statutes, chapter 325M.
1.5BE IT ENACTED BY THE LEGISLATURE OF THE STATE OF MINNESOTA:
1.6Section 1. new text begin [325M.39] TITLE. new text end
1.7 new text begin Sections 325M.40 to 325M.42 may be cited as the "Responsible Artificial Intelligence new text end
1.8 new text begin Safety and Education Act" or the "RAISE Act." new text end
1.9Sec. 2. new text begin [325M.40] DEFINITIONS. new text end
1.10 new text begin (a) For the purposes of sections 325M.40 to 325M.42, the following terms have the new text end
1.11 new text begin meanings given. new text end
1.12 new text begin (b) "Artificial intelligence" means a machine-based system that: (1) is able to, for a given new text end
1.13 new text begin set of human-defined objectives, make predictions, recommendations, or decisions new text end
1.14 new text begin influencing real or virtual environments; and (2) uses machine- and human-based inputs to new text end
1.15 new text begin perceive real and virtual environments, abstract the perceptions into models through analysis new text end
1.16 new text begin in an automated manner, and use model inference to formulate options for information or new text end
1.17 new text begin action. new text end
1.18 new text begin (c) "Artificial intelligence model" means an information system or component of an new text end
1.19 new text begin information system that implements artificial intelligence technology and uses computational, new text end
1.20 new text begin statistical, or machine-learning techniques to produce outputs from a given set of inputs. new text end
1.21 new text begin (d) "Critical harm" means the death, serious physical injury, or mental injury of 25 or new text end
1.22 new text begin more people, or at least $1,000,000 of damages to rights in money or property, caused or new text end
2.1 new text begin materially enabled by a developer's use, storage, or release of an artificial intelligence model new text end
2.2 new text begin that is the result of: new text end
2.3 new text begin (1) the creation or use of a chemical, biological, radiological, or nuclear weapon; or new text end
2.4 new text begin (2) conduct that with no meaningful human intervention would, if committed by a human, new text end
2.5 new text begin constitute a crime that requires intent, recklessness, or gross negligence, or the solicitation new text end
2.6 new text begin or aiding and abetting of a crime that requires intent, recklessness, or gross negligence. new text end
2.7 new text begin (e) "Developer" means a person that has trained at least one artificial intelligence model. new text end
2.8 new text begin (f) "Safety and security protocol" means a documented technical and organizational new text end
2.9 new text begin protocol that: new text end
2.10 new text begin (1) describes reasonable protections and procedures that, if successfully implemented, new text end
2.11 new text begin appropriately reduce the risk of critical harm; new text end
2.12 new text begin (2) describes reasonable administrative, technical, and physical cybersecurity protections new text end
2.13 new text begin for artificial intelligence models within the developer's control that, if successfully new text end
2.14 new text begin implemented, appropriately reduce the risk of unauthorized access to or misuse of the new text end
2.15 new text begin artificial intelligence models leading to critical harm, including by sophisticated actors; new text end
2.16 new text begin (3) describes in detail the testing procedure to evaluate whether the artificial intelligence new text end
2.17 new text begin model (i) poses an unreasonable risk of critical harm, (ii) could evade the artificial new text end
2.18 new text begin intelligence model's developer's or user's control, or (iii) could be misused, modified, new text end
2.19 new text begin executed with increased computational resources, combined with other software, or used new text end
2.20 new text begin to create another artificial intelligence model in a manner that increases the risk of critical new text end
2.21 new text begin harm; new text end
2.22 new text begin (4) enables the developer or third party to comply with the requirements of sections new text end
2.23 new text begin 325M.40 to 325M.42; and new text end
2.24 new text begin (5) designates senior personnel to be responsible for ensuring compliance. new text end
2.25 new text begin (g) "Safety incident" means a known incident of critical harm or one of the following new text end
2.26 new text begin that provides demonstrable evidence of an increased risk of critical harm: new text end
2.27 new text begin (1) an artificial intelligence model autonomously engages in behavior other than at the new text end
2.28 new text begin request of a user; new text end
2.29 new text begin (2) theft, misappropriation, malicious use, inadvertent release, unauthorized access, or new text end
2.30 new text begin escape of an artificial intelligence model's model weights; or new text end
2.31 new text begin (3) unauthorized use of an artificial intelligence model. new text end
3.1Sec. 3. new text begin [325M.41] TRANSPARENCY REQUIREMENTS. new text end
3.2 new text begin Subdivision 1. Developer requirements. Before deploying an artificial intelligence new text end
3.3 new text begin model, a developer must: new text end
3.4 new text begin (1) implement a written safety and security protocol; new text end
3.5 new text begin (2) retain an unredacted copy of the safety and security protocol, including records and new text end
3.6 new text begin dates of updates or revisions, for the entire period of time an artificial intelligence model new text end
3.7 new text begin is deployed, plus five years; new text end
3.8 new text begin (3) conspicuously publish a copy of the safety and security protocol with appropriate new text end
3.9 new text begin redactions, and transmit a copy of the redacted safety and security protocol to the attorney new text end
3.10 new text begin general; new text end
3.11 new text begin (4) grant the attorney general access to the safety and security protocol with redactions new text end
3.12 new text begin only to the extent required by federal law, if the attorney general requests access; new text end
3.13 new text begin (5) record and retain information on the specific tests and test results used in any new text end
3.14 new text begin assessment of the artificial intelligence model required under this section or by the developer's new text end
3.15 new text begin safety and security protocol that provides sufficient detail for third parties to replicate the new text end
3.16 new text begin testing procedure for the entire period of time an artificial intelligence model is deployed, new text end
3.17 new text begin plus five years; and new text end
3.18 new text begin (6) implement appropriate safeguards to prevent unreasonable risk of critical harm. new text end
3.19 new text begin Subd. 2. Prohibition. A developer must not deploy an artificial intelligence model if new text end
3.20 new text begin doing so creates an unreasonable risk of critical harm. new text end
3.21 new text begin Subd. 3. Annual review. (a) A developer must (1) conduct an annual review of the new text end
3.22 new text begin safety and security protocol required under this section to account for changes to the new text end
3.23 new text begin capabilities of the artificial intelligence model and industry best practices; and (2) modify new text end
3.24 new text begin the safety and security protocol. new text end
3.25 new text begin (b) If a material modification is made to the safety and security protocol, the developer new text end
3.26 new text begin must publish the safety and security protocol in the same manner required under subdivision new text end
3.27 new text begin 1, clause (3). new text end
3.28 new text begin Subd. 4. Safety incident disclosure. A developer must disclose each safety incident new text end
3.29 new text begin affecting the artificial intelligence model to the attorney general within 72 hours of the date new text end
3.30 new text begin the developer learns of the safety incident or within 72 hours of the date the developer learns new text end
3.31 new text begin sufficient facts to establish a reasonable belief that a safety incident has occurred. The new text end
3.32 new text begin disclosure must include: new text end
4.1 new text begin (1) the date of the safety incident; new text end
4.2 new text begin (2) the reasons the safety incident qualifies as a safety incident as defined in this section; new text end
4.3 new text begin and new text end
4.4 new text begin (3) a short statement describing in plain language the safety incident. new text end
4.5 new text begin Subd. 5. False or materially misleading statements. A developer must not knowingly new text end
4.6 new text begin make false or materially misleading statements or omissions in or regarding documents new text end
4.7 new text begin produced under this section. new text end
4.8Sec. 4. new text begin [325M.42] ENFORCEMENT; PRIVATE RIGHT OF ACTION. new text end
4.9 new text begin Subdivision 1. Attorney general. The attorney general may bring a civil action for a new text end
4.10 new text begin violation of section 325M.41 and recover, based on severity of the violation: new text end
4.11 new text begin (1) a civil penalty in an amount not exceeding $10,000,000 for a first violation and in new text end
4.12 new text begin an amount not exceeding $30,000,000 for any subsequent violation; and new text end
4.13 new text begin (2) injunctive or declaratory relief. new text end
4.14 new text begin Subd. 2. Private right of action. A person injured by a violation of this section may new text end
4.15 new text begin bring a civil action to recover damages, costs, and disbursements, including reasonable new text end
4.16 new text begin attorney fees, and receive other equitable relief as determined by the court. 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