North Dakota AI Laws and Regulation (2026)

North Dakota regulates AI through four laws enacted during its 2025 legislative session. HB 1167 requires a disclaimer on political content that uses AI to visually or audibly impersonate a human, HB 1351 addresses nonconsensual sexually expressive images including computer-generated ones, HB 1386 criminalizes computer-generated child sexual abuse material, and SB 2280 requires a licensed physician or licensed pharmacist, rather than software alone, to issue a health insurance prior authorization denial. A separate bill, HB 1320, would have broadly criminalized fraudulent deepfakes but failed 17-69 on the House floor on January 21, 2025, and never became law.

Overview of North Dakota AI Laws

North Dakota took decisive action on artificial intelligence during its 2025 legislative session, enacting multiple laws addressing deepfakes, political AI disclosures, health insurance decisions, and child protection. Because North Dakota holds legislative sessions only in odd-numbered years, the 2025 session represented the state's first opportunity to address the rapid growth of AI technology through legislation.

Governor Kelly Armstrong signed four relevant bills into law during the session, covering political ad disclosures, nonconsensual synthetic intimate images, computer-generated child sexual abuse material, and health insurance prior authorization through Senate Bill 2280. Only HB 1167 actually uses the term "artificial intelligence." HB 1351 and HB 1386 reach AI-generated material through broader "computer-generated" language, and SB 2280 reaches it indirectly by requiring a licensed human decisionmaker. A fifth bill targeting fraudulent deepfakes generally, HB 1320, failed on the House floor 17-69 on January 21, 2025, and never became law.

The state also established official AI guidelines for state agencies through the North Dakota Information Technology department, creating a governance framework for how state government uses AI tools.

This article covers North Dakota's enacted AI legislation, deepfake regulations, health insurance rules, and state government AI governance. This information is current as of March 2026, but you should consult an attorney for advice specific to your situation.

AI in Political Communications: HB 1167

House Bill 1167 addresses the use of artificial intelligence in political advertising. The law, which Governor Armstrong signed on April 11, 2025 after the Senate's final passage on March 18, 2025, creates a new section in Chapter 16.1-10 of the North Dakota Century Code (Corrupt Practices) requiring disclosure when AI is used to create certain political content.

Disclosure Requirements

The duty is narrower than it is often described. Under NDCC 16.1-10-04.2(1), the disclaimer is required only where the content was created in whole or in part with the use of artificial intelligence to visually or audibly impersonate a human. Content meeting that description must prominently state the disclaimer, which the statute sets in capital letters:

THIS CONTENT GENERATED BY ARTIFICIAL INTELLIGENCE.

The requirement reaches any action taken for a political purpose, including communication and political advertising in support of or opposition to a candidate, political committee, or political party, and content promoting the passage or defeat of initiated or referred measures or petitions. In practice that covers:

• Television and video advertisements
• Radio and audio advertisements
• Print materials and mailers
• Digital and online advertisements
• Social media political content

What the Disclosure Rule Does Not Cover

Subsection 2 creates an express exception. The section does not apply to content that uses artificial intelligence solely for text generation, grammar correction, spelling checks, stylistic editing, or enhancing existing content, without creating a new impersonation of a human likeness or voice. A campaign that drafts a press release with a chatbot, or runs a mailer through an AI grammar checker, does not trigger the disclaimer.

The statute's own definition reinforces that line. Subsection 3 defines artificial intelligence as a machine-based system capable of creating images, graphics, videos, audio, text, and other digital content based on human-defined objectives or data patterns, and expressly excludes systems programmed with rules or tools solely designed to assist with grammar, spelling, or word suggestions without generating human likeness or voice.

Purpose and Context

North Dakota lawmakers introduced HB 1167 in response to growing concerns about AI-generated political content that could mislead voters. The bill recognizes that as AI tools become more sophisticated, voters need transparency about whether a candidate they see or hear in an advertisement was actually depicted or synthesized.

The law places North Dakota among the 28 states that have enacted laws related to deepfakes used in political communications as of early 2026.

Fraudulent Deepfakes: HB 1320 Failed

House Bill 1320 would have created a broad prohibition against the fraudulent use of deepfake videos and images by adding a new section to Chapter 12.1-31 of the North Dakota Century Code (Miscellaneous Offenses). The bill did not become law.

What Happened to the Bill

HB 1320 was introduced on January 13, 2025, and referred to the House Judiciary Committee. On its second reading on January 21, 2025, the House rejected the bill 17 to 69, and it never advanced to the Senate or reached the Governor's desk.

What the Bill Would Have Prohibited

Had it passed, HB 1320 would have made it illegal to create, possess, or release deepfake videos and images without the consent of the person featured, defining a "deepfake video or image" as any digitally altered or AI-generated content that depicts an individual's likeness or voice without their consent, with the intent to deceive. Violations would have been a Class A misdemeanor, North Dakota's most serious misdemeanor classification, carrying up to 360 days imprisonment and a $3,000 fine.

Current Gap in North Dakota Law

North Dakota does not have a general criminal statute targeting fraudulent deepfakes outside the narrower categories covered by HB 1351 (nonconsensual sexually expressive images) and HB 1386 (computer-generated child sexual abuse material). A person who creates a fraudulent, non-sexual deepfake of another person, for example for financial fraud or reputational harm, would need to be prosecuted, if at all, under other existing fraud, harassment, or identity-related statutes rather than a deepfake-specific law.

Legislative History

Representative Josh Christy, R-Fargo, was the prime sponsor of HB 1320. He argued that deepfakes pose a threat to North Dakotans because it has become increasingly difficult to determine what is real and what is fake, but the bill did not persuade enough of his House colleagues to advance it.

Nonconsensual Synthetic Intimate Images: HB 1351

House Bill 1351 targets the creation and distribution of sexually explicit deepfakes without consent. Governor Armstrong signed the bill into law, and it took effect on August 1, 2025.

Criminal Penalties

The statute is narrower than a general ban on nonconsensual images. NDCC 12.1-27.1-03.3 defines two offenses, each requiring knowledge of the image's character and content.

A class A misdemeanor is committed by a person who either:

• Surreptitiously creates a sexually expressive image, or willfully possesses one that was surreptitiously created, without written consent from each individual who has a reasonable expectation of privacy in the image; or
• Distributes or publishes the image, electronically or otherwise, with intent to cause emotional harm or humiliation to a depicted individual who has a reasonable expectation of privacy in it, or after being given notice by that individual, or the individual's parent or guardian, that they do not consent to distribution or publication.

A class B misdemeanor is committed by a person who acquires and knowingly distributes a sexually expressive image that was created without the consent of the subject.

Merely possessing an image that was created consensually is not an offense under this section. The offense turns on how the image was made, or on the intent or notice accompanying its distribution.

The definition is what brings deepfakes inside the statute. NDCC 12.1-27.1-01(13) defines a "sexually expressive image" as a real, altered, or computer-generated photograph or visual representation that exhibits a nude or partially denuded human figure or sexual conduct, so a purely AI-generated image falls squarely within it.

Civil Remedies

Beyond criminal penalties, HB 1351 provides civil remedies for victims. The cause of action is not automatic: under NDCC 12.1-27.1-03.3(6), an identifiable depicted individual who suffers harm may sue only if the defendant produced, possessed, distributed, promoted, advertised, sold, exhibited, broadcast, or transmitted the image for the purpose of sexual arousal, sexual gratification, humiliation, degradation, or monetary or commercial gain.

A prevailing plaintiff recovers the greater of actual damages or statutory damages, not both, along with disgorgement of the defendant's gain and exemplary damages. The court may also protect the plaintiff's identity by ordering identifying characteristics redacted from filings.

HB 1320 Did Not Become Law

A companion bill, HB 1320, would have broadly criminalized fraudulent deepfakes but failed 17-69 on the House floor on January 21, 2025, and never became law. HB 1351 remains the only enacted North Dakota statute targeting synthetic intimate images specifically; a person who creates a fraudulent deepfake outside that sexually explicit category cannot be charged under a North Dakota deepfake-specific statute.

AI-Generated Child Sexual Abuse Material: HB 1386

House Bill 1386 addresses one of the most serious AI-related crimes: the creation and possession of computer-generated child sexual abuse material (CSAM). The law was introduced by Representative Josh Christy, R-Fargo.

Why This Law Was Needed

The number of reports of child sexual abuse materials in North Dakota has tripled since 2020, and state experts link the spike in part to realistic AI-generated content. Before HB 1386, investigators encountered AI-generated child abuse cases that could not be effectively prosecuted under existing law because the content did not depict real children.

Criminal Penalties

NDCC 12.1-27.2-04.1 establishes a two-tier penalty structure. Base possession is a Class C felony, and the offense rises to a Class B felony on any one of five statutory factors:

The statute also shields intermediaries. It does not impose liability on an interactive computer service as defined in 47 U.S.C. 230, or on an internet service provider, cloud computer provider, cybersecurity provider, or communication service provider, for content provided by another person.

Key Definitions

The law addresses the challenge of determining the "age" of fictitious AI-generated children. Because the content may involve entirely fabricated individuals, the law specifies that the ages of depicted children can be determined by secondary sex characteristics shown in the content. This approach allows prosecutors to pursue cases even when no real child was depicted.

Interaction with Federal Law

HB 1386 complements federal law rather than duplicating it. Section 2256 of Title 18 of the United States Code is the definitions section for the federal child exploitation chapter, and its definition of child pornography reaches visual depictions produced by computer as well as those involving actual minors. The operative federal prohibitions on producing, distributing, receiving, and possessing that material are 18 U.S.C. 2252 and 2252A, not the definitions section. State-level prosecution under HB 1386 provides an additional tool for law enforcement, particularly for cases that may not meet federal prosecution thresholds.

Health Insurance Prior Authorization: SB 2280

Senate Bill 2280 is frequently described as North Dakota's health care AI law, but that label does not match the enacted text. Governor Kelly Armstrong signed the bill on April 21, 2025 according to the Legislative Assembly's own action record, which reported the signing two days later on April 23 (the date carried by most news coverage). The Act took effect January 1, 2026.

What the Law Actually Says

SB 2280 creates Chapter 26.1-36.12 of the North Dakota Century Code, titled "Prior Authorization for Health Insurance." The phrases "artificial intelligence" and "algorithm" do not appear anywhere in the enrolled bill or in the codified chapter. There is no statutory prohibition on an insurer using AI, and no statutory definition of it.

What the chapter does instead is name the humans who must sign off. Under NDCC 26.1-36.12-03, a prior authorization review organization must ensure that all adverse determinations are made by a licensed physician or licensed pharmacist. The reviewing individual must have experience treating patients with the condition or illness for which the health care service is being requested, and must make the determination under the clinical direction of one of the organization's medical directors.

Because a denial must come from one of those licensed professionals, an insurer cannot issue a prior authorization denial through software alone. That is a practical consequence of a human-review mandate, not an AI ban, and it is why news coverage framed the bill as a check on AI in health care. The law does not bar an insurer from using AI to assist a qualified human reviewer.

Decision Timelines

The chapter sets deadlines for prior authorization decisions. Both of the main clocks run from the point the insurer has what it needs, not from the date of the original request, which is a meaningful difference for patients waiting on a decision:

If the provider or enrollee fails to supply the necessary information within fourteen calendar days of a written request for it, the review organization may make an adverse determination.

The default-approval rule sits in its own section. Under NDCC 26.1-36.12-13, if a prior authorization review organization fails to comply with the deadlines and other requirements of the chapter, any health care services subject to review are automatically deemed authorized.

Emergency care sits outside the prior authorization system entirely. Section 26.1-36.12-07 bars an insurer from requiring prior authorization for prehospital transportation or for emergency health care services for an emergency medical condition, and gives the enrollee or provider at least two business days after an emergency admission to notify the insurer.

Who Reviews a Denial

The chapter draws a line between the person who makes an adverse determination and the person who reviews it on appeal, and the qualification standards differ:

• Adverse determinations (26.1-36.12-03): a licensed physician or licensed pharmacist who has experience treating patients with the condition at issue, acting under a medical director's clinical direction.
• Appeals (26.1-36.12-04): a physician holding a valid nonrestricted license to practice medicine, in active practice in the same or similar specialty as the physician who typically manages the condition for at least five consecutive years, knowledgeable about and experienced in providing the service under appeal, not previously involved in making the adverse determination, and not receiving a financial incentive based on the number of adverse determinations made. An appeal involving a prescription drug may be reviewed by a licensed pharmacist or physician competent to evaluate the specific clinical issues.

The specialty-matching requirement therefore applies to appeal reviewers, not to the physician or pharmacist who issues the original denial.

The chapter also imposes transparency duties. Prior authorization requirements and restrictions, including the written clinical criteria, must be readily accessible on the organization's website in plain and ordinary language a layperson can understand. A new or amended requirement cannot be implemented until the website has been updated and contracted providers have received written notice at least sixty days beforehand.

Legislative Support

SB 2280 passed the North Dakota legislature with near-unanimous bipartisan support. The enrolled bill records a Senate vote of 43 yeas to 3 nays and a House vote of 93 yeas to 0 nays, reflecting broad agreement that a licensed clinician, rather than an automated process, should stand behind a coverage denial.

Impact on Healthcare

Healthcare organizations in North Dakota praised the law as a significant reform. Prior authorization has long been a source of friction between healthcare providers and insurance companies, and the increasing use of automated tools to process these decisions has amplified concerns about patients receiving denials without meaningful human review.

State Government AI Guidelines

The North Dakota Information Technology department (NDIT) has published official AI guidelines for state government use. These guidelines provide a governance framework that applies to all executive branch state agencies, including the University Systems Office.

Scope and Purpose

The NDIT AI guidelines outline best practices for:

• Secure use of AI technologies
• Privacy protection when using AI systems
• Ethical considerations in AI deployment
• Responsible innovation in government AI applications

State agencies outside NDIT's direct scope are encouraged to adopt the guidelines or use them as a framework for developing their own AI policies.

Key Definitions

The guidelines define three core AI concepts for state employees:

• Artificial Intelligence (AI): A field in computer science that focuses on independent decisions based on supervised and unsupervised learning
• Machine Learning (ML): A subfield of AI focused on algorithms and statistical models for independent decisions, while still requiring human guidance
• Large Language Models (LLMs): AI systems trained on large text datasets to understand existing content and generate original content

Privacy Considerations

The guidelines highlight important privacy risks associated with AI use in government. Data entered into public AI services is not secure, and public AI services may incorporate input data into their learning models. This means sensitive government data could potentially be exposed as output to other users if entered into public AI tools.

State employees are advised to avoid entering sensitive or confidential information into public AI services and to use approved, secure AI tools for government work.

AI in Employment

North Dakota has not enacted specific legislation governing the use of AI in employment and hiring decisions. Employers in the state using AI-powered hiring tools, resume screening systems, or automated evaluation processes must comply with existing federal anti-discrimination laws.

Federal Protections

Without state-specific AI employment legislation, North Dakota employers are governed by:

• Title VII of the Civil Rights Act prohibiting employment discrimination
• The Americans with Disabilities Act protecting candidates with disabilities from AI screening bias
• The Age Discrimination in Employment Act protecting workers over 40
• Equal Employment Opportunity Commission guidance on AI in employment decisions

Future Considerations

As AI hiring tools become more prevalent, North Dakota may consider employment-specific AI legislation in future legislative sessions. The state's next regular legislative session is scheduled for 2027, when lawmakers will have the opportunity to address any emerging concerns about AI in the workplace.

Federal AI Policy and North Dakota

Executive Order 14365

President Trump's Executive Order 14365, signed December 11, 2025, establishes federal AI policy that intersects with state regulatory efforts. The order creates mechanisms for reviewing state AI laws and conditions certain federal funding on states' regulatory approaches.

Impact on North Dakota Laws

North Dakota's enacted laws interact with the federal framework in several ways:

Deepfake laws (HB 1351, HB 1386): These laws address criminal conduct and child safety, areas where states have traditional authority. The child safety provisions of HB 1386 in particular fall within explicit federal carve-outs. (A broader fraudulent-deepfake bill, HB 1320, failed in the House and is not law.)

Political AI disclosure (HB 1167): Election regulation is primarily a state function, though federal preemption concerns could arise if the disclosure requirements are deemed to burden AI development. HB 1167 is the only one of the four laws that regulates AI by name.

Health insurance prior authorization (SB 2280): Insurance regulation has traditionally been a state domain under the McCarran-Ferguson Act, providing strong legal footing. Because SB 2280 regulates who may issue a coverage denial rather than the technology used to reach it, it presents little surface for a preemption challenge aimed at AI regulation.

State government AI guidelines (NDIT): Guidelines governing state agency AI use fall squarely within states' authority over their own government operations.

Comparison: North Dakota's AI Approach

North Dakota's 2025 legislative session produced one of the more productive AI-adjacent legislative outputs of any state that year. The state's approach is notable for several reasons:

Layered rather than omnibus: By enacting four separate bills covering political ad disclosure, nonconsensual sexually expressive images, computer-generated child sexual abuse material, and who may issue a health insurance denial, North Dakota built a layered framework rather than a single AI statute. A fifth bill targeting fraudulent deepfakes generally, HB 1320, failed in the House and left that category of AI misuse without a dedicated state criminal statute.

Indirect regulation: Three of the four laws never use the phrase "artificial intelligence." They reach AI-generated conduct through technology-neutral drafting, either by covering "computer-generated" images or by requiring a licensed human to make a decision. That drafting choice makes the laws harder to outrun as the technology changes, but it also means North Dakota has no general-purpose AI statute.

Health insurance focus: SB 2280 places North Dakota among the early states to constrain automated health insurance denials, a trend gaining momentum nationally, though it does so through a human-review mandate rather than an express AI restriction.

Biennial session constraint: Because North Dakota meets only every two years, the 2025 session represented the state's only opportunity to legislate on AI until 2027. This constraint may have contributed to the volume of legislation produced.

More North Dakota Laws

Explore other North Dakota law topics on Recording Law:

• North Dakota Recording Laws
• North Dakota Data Privacy Laws
• North Dakota Background Check Laws
• North Dakota Biometric Privacy Laws
• North Dakota Data Breach Notification Laws