What is AI regulation in New Hampshire?
AI regulation: countries and regions
New Hampshire has no single AI law but has enacted a few narrow rules. The state makes certain harmful deepfake uses a felony and lets victims sue over fraudulent AI imagery. It requires political campaigns to label AI-generated ads near elections. A 2026 law prohibits chatbot operators from knowingly encouraging minors to self-harm or commit crimes. State agencies using AI must have human review and tell users they are interacting with AI.
Reviewed by Jackie, Head of Learning & Development, Levellers · Last reviewed 18 June 2026
What this means
New Hampshire's approach to AI is largely piecemeal. Instead of a broad AI framework, the state has added AI rules to existing laws. For example, a law effective 2025 criminalizes creating deepfake images or audio designed to harass or defraud someone, and gives victims a civil right to sue over such misuse. In 2024 it passed a law requiring disclosure if a political advertisement uses AI-generated synthetic media near elections.
Another law (effective 2026) expands child protection laws: owners of AI chat programs can be prosecuted if they knowingly use the chatbot to entice a child to do something dangerous (like self-harm or violence). New Hampshire also set rules for its own agencies: RSA Chapter 5-D (effective 2024) requires human oversight of irreversible AI decisions and that users be informed when they interact with an AI system. Outside of these specific rules, general state laws (such as consumer protection or data privacy statutes) apply to AI just as they do to other business practices.
Why it matters
For organisations in New Hampshire, these AI rules mean compliance in specific areas. A company making or selling AI chatbots should avoid designs that could encourage a minor to harm themselves or others, or risk criminal penalties and civil suits. Campaigns and political advertisers in the state must prominently label AI-generated or manipulated content in ads during the election period, or face penalties. State contractors using AI systems for decisions (such as in healthcare or law enforcement) need human oversight and user notice under RSA 5-D. Overall, even though New Hampshire's AI rules are narrow, failing to meet these specific duties (disclosure, oversight, etc.) could lead to fines, lawsuits or enforcement actions by the state Attorney General.
How it works
Deepfakes and Fraudulent AI Content
New Hampshire amended its criminal and civil codes to cover AI "deepfakes." Under RSA 638:26-a, it is a class B felony to knowingly create or share a deepfake (audio/video/media altered to impersonate someone) intended to embarrass, harass, defraud, or harm a person. Similarly, RSA 507:8-j lets an individual sue anyone who used a deepfake to harm them (for example to extort or damage reputation). These provisions target deliberate misuse of AI-generated or AI-altered media. Courts can impose prison time for the criminal offense, and victims can seek damages in civil court if they suffer reputational or financial harm. Exemptions cover bona fide news reporting and parody. Enforcement is by New Hampshire's Attorney General and local prosecutors; victims can also sue private actors under the civil section.
AI in Political Advertising
In 2024 New Hampshire added a new election-law disclosure. RSA 664:14-c makes it unlawful for any person or group to distribute an AI-generated or manipulated image, video, or audio of a candidate or party within 90 days of a state or local election unless it includes a clear label. The law defines "synthetic media" and "deepfake" and requires any AI content in ads to be captioned with text such as "This image/video/audio has been generated by artificial intelligence". The campaign must use the word "Image," "Video," or "Audio" in the disclosure. If the ad is audio-only, the message must be spoken audibly at intervals. Candidates or officials depicted can seek injunctions or damages if these rules are violated. The Attorney General's office enforces this section, and violation can lead to removal of the ads and fines.
Chatbot and Child Safety Law
From January 1, 2026, New Hampshire treats harmful chatbot content as child endangerment. A 2025 law (RSA 639:3 III-a) says that an owner or operator of an AI chat service (including text chatbots, "character AI," or similar) who "direct[s]" a child's conversation with intent to induce the child to self-harm, commit violence, engage in sexual conduct, or other listed acts commits a felony. This requires "knowing" intent: the operator must be aware that the chatbot's instructions are meant to cause such behavior. The law specifically targets owners of AI chat platforms whose sole purpose is open-ended communication with children. Penalties are severe: violating this provision is a class B felony. This provision does not impose requirements on other AI uses, but strictly prohibits predatory AI chat content targeting children.
State Agency AI Use (RSA 5-D)
New Hampshire also regulates AI use by its own agencies. RSA Chapter 5-D (effective July 1, 2024) governs any state department's AI deployment. Key rules include: if an AI system's output cannot be reversed (for example an automatic order or decision), a human supervisor must review it before it takes effect. Any generative AI content released by a state agency without human review must be labeled as AI-generated. And if a person is interacting (directly or via a service) with an AI system, the agency must inform them they are using AI. The DoIT (Dept. of Information Technology) is charged with overseeing compliance: each agency must audit its AI tools within 9 months, remove any prohibited uses, and annually report to the governor on its AI systems and compliance.
Enforcement and Oversight
Enforcement of AI-related rules falls mostly to existing authorities. The Attorney General enforces criminal and consumer laws, including the deepfake and election provisions. Victims of illegal AI deepfakes or election-law violations can bring lawsuits, while class B felonies (deepfakes, child endangerment by chatbot) are prosecuted like any other serious crime. The state's legislative information indicates fines and prison terms for violations. State agencies themselves must follow RSA 5-D or face administrative consequences. There is no standalone "AI enforcement agency"; responsibilities lie with the AG, DoIT, and the judiciary.
Examples
A campaign consultant in Manchester produces a video ad using AI to depict a rival politician saying things they never did. Under RSA 664:14-c, the consultant must add a visible caption, e.g. "This video has been generated by artificial intelligence". If the ad airs without this disclosure, the state could seek an injunction and the candidate could sue for damages.
A tech startup offers an AI-powered chatbot for anxiety support. If the company's chatbot persuades a minor to engage in disordered eating or self-harm, and the operator knew that was the intent, RSA 639:3 III-a would make that operator criminally liable. The startup must program safeguards and monitoring to prevent any such outcome.
A New Hampshire health department uses an AI tool to prioritize patient referrals. Because this decision is consequential, the official policy in RSA 5-D:4 requires a human manager to review the AI's output before acting. If the department rolled out the AI without human oversight, it would violate the state code.
An individual finds a deepfake image of themselves online that falsely portrays them committing a crime. Under RSA 638:26-a and 507:8-j, they could notify law enforcement and also sue the creator of the deepfake for damages. Law enforcement could prosecute the creator for creating the illegal deepfake.
Common misunderstandings
"NH has a broad AI law": No. New Hampshire hasn't passed an omnibus AI bill. It regulates only certain AI uses (deepfakes, child chatbots, election ads) through specific statutes.
"Private chatbots must always identify themselves as AI": Not under state law for private businesses. The RSA 5-D disclosure requirement applies to state agency services, not to all vendors. Private firms must disclose in political ads (and comply with general deception laws), but ordinary consumer AI tools have no special labeling law.
"Any AI decision must be reviewed by a human in NH": Only for certain state agency uses. RSA 5-D mandates human review of irreversible AI decisions in the public sector. There's no general NH law requiring human review of AI tools in private industry.
"All deepfakes are illegal": NH law targets deepfakes used for fraud, harassment or defamation. Satire or news use is exempt. Creating a harmless AI-generated parody isn't automatically a crime, but harmful or deceptive deepfakes violate RSA 638:26-a and possibly RSA 507:8-j.
"AI regulations override federal law": State laws like RSA 664:14-c cover state/local elections, but federal law also governs national races and media. Similarly, NH criminalizes deepfakes locally, but federal prosecutions (e.g. for cybercrime) remain separate.
Risks and boundaries
New Hampshire's AI rules are narrowly focused. They do not ban or broadly regulate AI development, algorithmic bias, or data usage - those matters remain largely unaddressed by state law. The state relies on existing frameworks, so most AI uses fall under general legal duties (like existing consumer protection, anti-discrimination laws, and federal regulations) unless a specific AI rule applies. For example, there is no New Hampshire privacy law giving individuals new data rights over AI training data.
Be aware that enforcement practices are still evolving. Some proposed bills (like a comprehensive "AI Council" law) were voted down or stalled, so future changes are possible but uncertain. Fines and punishments under these laws can be substantial (deepfakes and child safety violations are felonies), but prosecutors must prove intent and knowledge. The election disclosure rule applies only in the run-up to elections, not to marketing year-round.
Finally, these laws apply to all who operate in New Hampshire or target its citizens: an out-of-state firm placing deepfake ads in NH might still be held liable under RSA 664:14-c or similar statutes. Conversely, federal and industry standards also apply in NH, so organizations should consider both sets of rules.
What to do next
Review AI content policies. Audit any AI-generated media (ads, social posts, chatbots) for compliance. Ensure political campaigns and marketing teams know to label AI-generated content per RSA 664:14-c.
Safeguard children's interactions. If your products or services use AI chat for minors, implement strict monitoring. Don't allow content that could be seen as encouraging self-harm, violence or illegal acts. Remember that the operator's knowledge matters - train your staff about the new child protection provision.
Institute oversight. For state agencies or contractors using AI systems for important decisions, put formal human review steps in place as RSA 5-D requires. Even if you are private sector, similar policies reduce risk and align with best practices in AI governance.
Monitor compliance obligations. Update terms of use or disclaimers to align with NH law. Track legislative sessions - a serious AI governance bill was proposed in 2025 and may resurface. Engage with trade groups or legal counsel to stay ahead of any new NH AI rules.
Document and train. Keep records of how AI systems are used or tested, especially if state-funded. Train developers and managers on NH's specific prohibitions. Clear internal guidelines on "what NH law allows or forbids" will help avoid inadvertent violations.
FAQs
Does New Hampshire have a general AI law or regulator?
No. New Hampshire has not created a standalone AI framework or regulator. Instead, it has added AI provisions to existing laws (criminal code, elections laws, etc.) covering specific uses. There is no state "AI commission" or data privacy board yet.
What penalties exist for illegal AI content in NH?
Creating a harmful deepfake or using a chatbot to endanger children are Class B felonies under state law. Penalties for a Class B felony can include prison time and fines. Election-law violations (AI ads without labels) can lead to court injunctions and civil damages. The Attorney General's office enforces these.
If I run a chatbot in NH, do I need to label it as AI?
Only state agencies are required by RSA 5-D to tell users they're interacting with AI. Private chatbots (even bots like ChatGPT) aren't broadly mandated to disclose. However, if your chatbot gives advice or instructions, be aware of other laws - for example, it must not knowingly encourage a minor to break the law or self-harm, per the child welfare law.
How do NH laws interact with federal or other states' rules?
New Hampshire's laws apply within the state. Campaign ads using AI still must follow federal rules too. If a firm outside NH targets New Hampshire voters or users, it can be liable under these state rules. But NH's rules do not override valid federal regulations; they operate alongside them.
Can a person sue if an AI deepfake damages their reputation?
Yes. RSA 507:8-j gives any person the right to sue someone who knowingly used a deepfake to harm them. The law defines harm broadly (embarrassing, defrauding, etc.), so victims can seek money damages for losses caused by the deepfake.
Are there any requirements for AI in healthcare or finance?
Not specifically in state law beyond general rules. New Hampshire's RSA 5-D focuses on government agencies, not private health or financial firms. Those sectors should follow federal and industry standards. The state insurance department has issued guidance on AI use by insurers, but private companies must also comply with federal laws like HIPAA or FTC regulations.
What happens if a website's AI gives someone bad advice?
There's no NH law directly for AI advice errors. Such cases would fall under usual legal theories (product liability, negligence, defamation, etc.). If the AI advice caused harm, general consumer protection or malpractice laws (not AI-specific statutes) would apply unless it violated a specific NH AI rule described above.
Is New Hampshire considering a comprehensive AI law?
A "Responsible AI Governance" bill (HB 1725 in 2026) was introduced to create an AI council, sandbox, and enforce rules similar to RSA 5-D but for all sectors, but it was voted Inexpedient to Legislate in early 2026. Future proposals may arise, but as of mid-2026, no broad law has been enacted.
Sources
New Hampshire State Legislature: RSA 5-D:4 Permitted Uses and Restrictions (Use of AI by State Agencies) - New Hampshire law requiring human oversight of irreversible AI decisions by state agencies and disclosure that content is AI-generated.
New Hampshire State Legislature: RSA 638:26-a Fraudulent Use of Deepfakes (New Hampshire Revised Statutes) - Defines "deepfake" and makes creation or distribution of malicious deepfakes a Class B felony.
New Hampshire State Legislature: RSA 507:8-j Civil Actions for Fraudulent Use of Deepfakes (New Hampshire Revised Statutes) - Provides a civil cause of action for individuals harmed by others' fraudulent use of deepfakes.
New Hampshire State Legislature: RSA 664:14-c Synthetic Media and Deceptive and Fraudulent Deepfakes (New Hampshire Revised Statutes) - Requires political ads to disclose AI-generated or manipulated synthetic media by labeling it as AI-generated.
New Hampshire State Legislature: RSA 639:3 Endangering Welfare of Child or Incompetent (New Hampshire Revised Statutes) - Amended to criminalize certain AI-chatbot actions that knowingly encourage minors to self-harm or commit crimes; such acts are Class B felonies.