AI Regulation Tracker / US state legislation

Wyoming's AI Crime Law Is Now In Force, Not a Bill

Wyoming HB0102 became House Enrolled Act No. 32, chapter 91, and took effect on July 1, 2026. It creates felony offenses for building or distributing an AI system designed to promote self-harm or to produce child pornography, criminalizes distributing nonconsensual synthetic sexual material of any identifiable person, bars the argument that the AI rather than the defendant committed the crime, and gives developers a limited civil immunity. The AI political-speech provision in the engrossed bill is not in the enacted law.

By Anthony Guerriero · The Leveraged Years Regulatory Research · Published March 7, 2026 · Last updated July 29, 2026 · State legislation

Key facts

Instrument: House Bill HB0102, "Protecting kids from deepfakes and exploitative images," enacted as House Enrolled Act No. 32, chapter 91, 2026 Budget Session
Authority: Wyoming Legislature. Criminal provisions enforced by Wyoming prosecutors
Jurisdiction: State of Wyoming, United States (sub-national)
Status: Enacted. The Legislature's bill record lists a signature date of March 7, 2026, no veto, and a last action of "Assigned Chapter Number 91." The posted enrolled-act PDF leaves its signature and date lines blank, so the March 7 date comes from the bill record, not that PDF
Bindingness: Binding criminal law in force. Creates felony offenses carrying prison terms and fines
Effective date: July 1, 2026, stated in Section 3 of the enrolled act as "This act is effective July 1, 2026."
Sections created: W.S. 1-1-143, 6-1-206, 6-4-307, 6-4-308 and 6-4-701. Section 2 amends definitions in W.S. 6-1-104(a), 6-4-301(a) and 6-4-303
Not enacted: Proposed W.S. 40-12-801, the AI political-speech prohibition, is in the engrossed bill and nowhere in the enrolled act
Primary Source: Wyoming Legislature, HB0102 enrolled act, wyoleg.gov
Workflow Impact: Teams shipping generative AI to Wyoming users are inside a live criminal statute and should map their systems against the intent and knowledge elements, not the bill text
Operational Next Step: Re-read any note written from the engrossed bill. Drop the moderation workstream built on proposed section 40-12-801 and add the adult provision at W.S. 6-4-307
Last Verified: July 29, 2026

This page previously called HB0102 an engrossed bill that was not current law. That was wrong in two directions at once. The measure passed and has been enforceable since July 1, 2026, and the single provision this page treated as the most interesting one never became law at all.

The enrolled act is the version that matters. Its Section 1 creates five statutes, Section 2 amends three definition sections, and Section 3 says in full: "This act is effective July 1, 2026." The bill record shows a signature date of March 7, 2026, no veto, and a final action assigning chapter number 91.

What does the enacted act actually criminalize?

Start with the rule that carries across every other section. New W.S. 6-1-206 provides that "Nothing in the Wyoming Criminal Code shall be construed to prohibit the application of this code against activity facilitated by an artificial intelligence system," and that "The use of an artificial intelligence system by a person to commit a criminal offense shall not be a defense to a criminal charge." That is the load-bearing provision: the rest of Wyoming's criminal code already reaches AI-assisted conduct without waiting for an AI-specific offense.

Then come the offenses. New W.S. 6-4-701, inside a newly created Article 7 headed Artificial Intelligence Offenses, makes it a felony to knowingly develop or distribute an AI system "specifically designed to promote self-harm" where the person acts with intent or knowledge that others will use it that way. The penalty is up to ten years, a fine not to exceed ten thousand dollars, or both. The section then carves out a list that matters more than the offense in practice: output produced solely from a user's own prompts without the developer's intent, bona fide school, museum or public library activity, law enforcement and judicial activity, licensed health care providers delivering legitimate medical or therapeutic services, and intermediaries carrying content provided by another person.

New W.S. 6-4-308 mirrors that structure for child pornography, making it a felony to knowingly develop or distribute an AI system specifically designed to create, distribute or promote child pornography or synthetic sexual material, with the same carve-outs. Section 2 then folds AI into the existing sexual-exploitation-of-children offense, adding using an AI system to generate child pornography and possessing child pornography known to be AI-generated.

The provision most likely to be missed is W.S. 6-4-307, and it is not limited to children. It criminalizes distributing or making available synthetic sexual material that realistically depicts "another identifiable person" without their knowledge or consent, where the person knew it was made without consent in circumstances carrying a reasonable expectation of privacy, and where the distribution causes or is intended to cause harm. It also reaches promoting such material on a platform knowing its character, and threatening to distribute it for a benefit. It is a felony carrying up to ten years, rising to twelve for a second conviction, and it removes two defenses: that the depicted person consented to the material's creation, and that they voluntarily sent sexual material to the defendant.

What happened to the political-speech section?

The engrossed bill created six sections. The enrolled act creates five. The one that did not survive is proposed W.S. 40-12-801, which would have prohibited interactive computer services from using AI to block, ban, demonetize, restrict or otherwise discriminate against a user based on political speech, enforced civilly by the Attorney General at up to ten thousand dollars per violation.

The string "40-12-801" appears in the engrossed text and appears zero times in the enrolled act. The only trace left is in the act's own long title, which still recites offenses concerning the use of AI "to promote self-harm or to censor political speech." A legislative title is not an operative provision, and a title that outlived the section it described does not create a duty. There is no Wyoming AI political-speech obligation in force.

The earlier version of this page called that section "the novel one" and told readers to run a moderation workstream against it. Anyone who did was building against text already removed. This page's address still contains "political-speech-crimes," an artifact of the bill it was written from, and is left unchanged so existing links keep working.

Engrossed bill vs enacted law
Provision	Engrossed HB0102	Enrolled Act No. 32, in force
AI is not a defense to a crime	Proposed W.S. 6-1-206	Enacted, W.S. 6-1-206
Limited developer civil immunity	Proposed W.S. 1-1-143	Enacted, W.S. 1-1-143
Nonconsensual synthetic sexual material, identifiable person	Proposed W.S. 6-4-307	Enacted, W.S. 6-4-307, up to 10 years, 12 on repeat
AI system built to produce child pornography	Proposed W.S. 6-4-308	Enacted, W.S. 6-4-308, up to 10 years
AI system built to promote self-harm	Proposed W.S. 6-4-701	Enacted, W.S. 6-4-701, up to 10 years
AI political-speech censorship by online services	Proposed W.S. 40-12-801, AG-enforced	Not enacted. Retained only in the title
Effective date	July 1, 2026 if enacted	July 1, 2026, in force

When are developers protected, and when are they not?

New W.S. 1-1-143 is the counterweight to the offenses, and it sits in the civil code rather than the criminal code. It provides that the developer of an AI system "shall not be liable for damages arising from the use of an artificial intelligence system when another person used the system with the intent to commit or engage in illegal or illicit activities or to cause harm."

The exception is the whole point. That immunity does not apply "if the artificial intelligence system was developed with the knowledge or intent that the primary purpose of the system would be for illegal or illicit activities." Wyoming has drawn the line at purpose. A general-purpose model a third party misuses sits inside the immunity. A system whose primary purpose the developer knew to be illicit sits outside it, and that same knowledge is close to the element the criminal sections require. The civil shield and the criminal exposure turn on one question, which is a reason to document design intent and not only usage policies.

Why should professionals outside Wyoming track it?

Because the drafting pattern is portable and the enacted version is narrower than the bill. Wyoming did not legislate against AI in general. It attached liability to building a system for a prohibited purpose, kept a safe harbour for prompt-driven output the developer did not intend, preserved the federal intermediary carve-outs by reference, and dropped the constitutionally contested moderation provision before it reached the governor. The parts that were cut tell you as much as the parts that passed.

The practical instruction is narrow. Find out whether anything in your organization was written against the engrossed bill, because that text now misleads in both directions. Add the adult nonconsensual provision. Retire the moderation workstream. Then treat design intent as a documented artifact, because under both the criminal sections and the civil immunity it is the fact that decides the outcome.

Source File

Primary: Wyoming Legislature, HB0102 enrolled act (House Enrolled Act No. 32), wyoleg.gov/2026/Enroll/HB0102.pdf
Primary (status, chapter, dates): Wyoming Legislature bill record for HB0102, wyoleg.gov/Legislation/2026/HB0102
Primary (superseded text, for the comparison): Wyoming Legislature, HB0102 engrossed text, wyoleg.gov/2026/Engross/HB0102.pdf
How to verify: In the enrolled PDF read Section 1 for the statutes created, W.S. 6-1-206 for the no-defense rule, W.S. 6-4-307 for the identifiable-person offense, W.S. 6-4-701 for the self-harm felony and its exclusions, W.S. 1-1-143 for the developer immunity, and Section 3 for the effective date. Then search it for "40-12-801" and confirm nothing returns. The bill record supplies the chapter and signature dates
Editorial note: Informational analysis, not legal advice

Last verified: July 29, 2026

Corrected July 29, 2026. This page previously described HB0102 as an engrossed bill that had not been enacted, and treated proposed section 40-12-801 on AI political-speech censorship as operative. Both statements were wrong. The act took effect July 1, 2026 as House Enrolled Act No. 32, chapter 91, and section 40-12-801 does not appear in the enacted text. This revision was written from the enrolled act and the Legislature's bill record. Coverage of W.S. 6-4-307 and the developer immunity at W.S. 1-1-143 has been added. The page address is unchanged.

Questions professionals are asking

Is Wyoming HB0102 actually law?

Yes. It was enacted as House Enrolled Act No. 32 and assigned chapter 91 of the 2026 session laws, and Section 3 of the enrolled act states that the act is effective July 1, 2026. The Legislature's bill record lists a signature date of March 7, 2026 and no veto. Any note describing it as a pending or engrossed bill is out of date.

Did Wyoming ban AI political-speech censorship?

No. That prohibition existed in the engrossed bill as proposed section 40-12-801, which would have been enforced civilly by the Attorney General. It does not appear anywhere in the enrolled act. The act's long title still mentions censoring political speech, but a title is not an operative provision, so there is no Wyoming duty governing AI content moderation.

Is the self-harm provision a crime, and what are the exceptions?

It is a crime. W.S. 6-4-701 makes knowingly developing or distributing an AI system specifically designed to promote self-harm a felony punishable by up to ten years, a fine up to ten thousand dollars, or both, where the person acts with intent or knowledge that others will use it that way. It does not apply to output produced solely from a user's own prompts without the developer's intent, to bona fide educational, museum or library activity, to law enforcement and judicial activity, to licensed clinical care, or to intermediaries carrying another person's content.

Does the law cover adults or only children?

Both. Alongside the child-specific offenses, W.S. 6-4-307 criminalizes distributing synthetic sexual material that realistically depicts any other identifiable person without their knowledge or consent, where the distribution causes or is intended to cause harm. It is a felony carrying up to ten years, and twelve for a second conviction. It also removes the defenses that the depicted person consented to the material's creation or voluntarily sent sexual material to the defendant.

Can a defendant argue that the AI committed the offense?

No. W.S. 6-1-206 provides that using an AI system to commit a criminal offense is not a defense to a criminal charge, and that nothing in the Wyoming Criminal Code prevents applying the code to activity facilitated by an AI system. Responsibility stays with the person rather than the tool.

Are AI developers liable when someone else misuses their system?

Usually not, but the protection is conditional. W.S. 1-1-143 says a developer is not liable for damages when another person used the system intending illegal activity or harm. That immunity is lost if the system was developed with knowledge or intent that its primary purpose would be illegal, which makes documented design intent the decisive fact under both the immunity and the criminal sections.