This article was written by Taylor Barkley and Bryce Chinault of the Abundance Institute, an SPN partner.

As states grapple with how to respond to AI, some are drawing sensible lines around harmful conduct. Others are starting to move toward something more troubling — requiring government permission to innovate.

To push back, states are increasingly considering “right to compute” laws. The “right to compute” is about protecting the ability of researchers, entrepreneurs, and innovators to develop and use AI tools without government requiring a permission slip to do so. The question at hand is simple: Will we allow AI innovation to remain generally open and lawful, or become increasingly permissioned and compliance-driven?

However, this is not just about large AI models or tech companies. It covers the devices and systems that modern life runs on, including smartphones, laptops, home servers, business software, algorithms, cloud systems, and other computing tools. The core principle is simple: Lawful computing should be presumed free, and the government should have to justify any restriction with a narrow and compelling reason.

As AI models and the power of computer chips advance, there will likely be increased calls for restrictions on what kinds of computers citizens can access. Restrictions on open-source AI models could prevent developers from sharing tools that are widely used for research and entrepreneurship. Limits on computing that could force individuals and businesses to rely exclusively on centralized cloud providers. Broad restrictions on algorithmic systems which could inadvertently affect everything from agricultural equipment to logistics software.

A right to compute provides a defense against all of these freedom and innovation constraining proposals which have already been made in the US.

The right to compute fits squarely within the American tradition of technological freedom. Americans have long assumed the freedom to own tools, experiment with them, and use them to build new things. From the printing press to the personal computer, technological progress has flourished because individuals were generally free to explore and innovate without first seeking permission from the state. Computing power is the modern equivalent of those tools. Protecting the right to compute ensures that the same tradition of experimentation and discovery continues in the digital age.

What Right to Compute Laws Do

A Right to Compute law generally does three things.

First, it protects lawful access to computational tools. That includes the right to own and use hardware, software, and algorithms for lawful purposes.

Second, it places a high burden on the government before restrictions may be imposed. In practice, that means regulators should need to show a compelling public interest and use a narrow approach rather than broad restrictions.

Third, it preserves safeguards where they are truly needed. The leading Right to Compute laws and proposals do not abolish safety rules. Instead, they pair broad protection for lawful computing with focused requirements for AI systems involved in critical infrastructure, often through risk management policies and human oversight requirements. Montana’s enacted law does exactly this, and Ohio’s pending bill follows the same general model. 

What Right to Compute Laws Don’t Do

Passing a so-called right to compute does not exempt developers and deployers from abiding by existing laws. A driver using Google Maps would not suddenly be exempt from a traffic violation because they were using this common AI powered tool nor would a would-be thief be excluded from criminal charges simply because his crime was carried out utilizing online platforms...

Federal and state level policymakers have made this perfectly clear. During the Biden administration, the Department of Justice, Federal Trade Commission, Consumer Financial Protection Bureau, and Equal Employment Opportunity Commission issued a joint statement clarifying how long-standing laws apply to the use of automated decision systems.

Connecticut Attorney General Willam Tong published a memorandum clarifying the same point, as have other state leaders like Massachusetts Attorney General Andrea Campbell.

The right to compute is not a right to break the law. It is an assurance that all of us are free to access computational resources and to use lawfully in creative ways that regulators cannot foresee; like when a bakery owner in Japan attempting to solve a problem at the checkout counter inadvertently helped save cancer patients.

What Does an Infringement Look Like?

A helpful way to understand the right to compute is to imagine what happens when the government treats ordinary or advanced computing as something people must ask permission to use.

Examples of what this shift toward a permission-first looks like in reality:

A family is told its home server or local AI system is too powerful to run without special approval. That would be an infringement. The state is no longer regulating harmful conduct. It is regulating lawful access to computing power itself. Proposals like California’s SB 1047 moved in this direction by imposing regulatory requirements on AI systems based on their scale and computational power—conditioning development and deployment not on demonstrated harm, but on the capabilities of the system itself. The computational thresholds in that specific bill were meant to target only the largest developers, but it would have set a permission slip precedent that would make it easier to regulate any computational threshold in the future.  

A startup cannot deploy software until it clears a vague licensing or approval process simply because the product uses advanced AI. That would be an infringement. The issue is not fraud, sabotage, or negligence. The issue is that the government has converted lawful computing into a permissioned activity. 

Well-intentioned proposals to limit misbehavior, like the Colorado AI Act, instead sweep small businesses into compliance. Such proposals would also be an infringement. Illegal behavior remains illegal. A right to compute would give Coloradans a defense when the state stacks additional requirements on individuals, small businesses, and others just because they are using software to complete tasks.

How States Are Protecting the Right to Compute

Several states have already moved legislation or constitutional language based on this idea. Together, they show that the right to compute is no longer just a theory. It is becoming a live state policy question. 

Montana

Status: Enacted, sponsored by State Sen. Daniel Zolnikov

Montana became the first state to enact a Right to Compute Act, thanks to the work of Tanner Avery and the team at Frontier Institute, when SB 212 was signed in April 2025. The law creates baseline protections for the ownership and use of computational resources and requires a risk management policy for critical infrastructure facilities controlled by an artificial intelligence system. In other words, Montana paired a broad liberty principle with targeted safeguards for high consequence uses. It passed with broad bipartisan support.

Official bill page: Montana SB 212

Ohio

Status: Introduced, sponsored by State Reps. Tex Fischer and Steve Demetriou

Ohio’s H.B. 392 would enact the Ohio Right to Compute Act. The bill would limit further regulation of certain computational systems and require risk management policies for AI controlled critical infrastructure, like electric grid management systems. The Ohio proposal matters because it would place a major industrial and technology state on record in favor of protecting lawful computational activity while still recognizing legitimate public safety concerns. Greg Lawson of the Buckeye Institute testified (10:50 mark) in support of the measure at a recent hearing.

Official bill page: Ohio H.B. 392

South Carolina

Status: Introduced, sponsored by State Reps. Brandon Guffey, Tommy Pope, and Melissa Oremus

South Carolina’s H. 4657 would enact a Right to Compute Act by adding a new chapter to state law. The bill, advocated by the Palmetto Promise Institute, would require risk management policies for critical infrastructure controlled by artificial intelligence systems and establish when private computational resources may be restricted. South Carolina’s proposal is important because it shows the idea spreading beyond the Mountain West and Midwest into the Southeast, where policymakers are also thinking about AI governance in terms of freedom, property, and lawful technological use.

Official bill page: South Carolina H. 4657

New Hampshire

Status: Introduced, sponsored by State Reps. Keith Ammon, Jeanine Notter, Vanessa Sheehan, Travis Corcoran, and Katy Peternel

New Hampshire lawmakers have introduced a statutory Right to Compute bill, H.B. 1124, in the 2026 session. Unlike CACR 6, described below, which proposed a constitutional amendment in 2025, H.B. 1124 would create a new chapter of state law called the Right to Compute Act.

The bill states that New Hampshire’s constitutional protections for property, free expression, and privacy also embody a fundamental right to own and use technological tools, including computational resources. It would prohibit government entities from restricting the private ownership or use of computational resources for lawful purposes unless the restriction is demonstrably necessary and narrowly tailored to fulfill a compelling government interest.

During a public hearing on the bill, Rep. Ammon provided what may have been a first in the nation moment when he had an AI tool provide direct audio testimony to the committee.

Official bill page: New Hampshire H.B. 1124

Status: Lapsed, introduced by State Reps. Keith Ammon, Bob Lynn, and Lex Berezhny

New Hampshire lawmakers proposed CACR 6, a constitutional amendment recognizing an individual right to use computation resources. The proposal would have protected the right of individuals to use computational resources, including devices and networks essential for computation. Although the proposal did not become law and was killed in the House, it remains an important example of how some lawmakers are thinking about this issue at the constitutional level rather than only through statute.

Official bill page: New Hampshire CACR 6

Key “Right to Compute” Policy Design Principles

As more states consider Right to Compute legislation, a few design principles stand out.

A good law should clearly protect lawful computing activity. It should say plainly that people and organizations may own and use computational tools for lawful purposes.

It should use a high standard for restrictions. If the government wants to limit access to computational resources, it should have to show a compelling interest and adopt a narrowly tailored response.

It should regulate harmful conduct rather than computational capacity as such. Existing law already addresses most of the issues raised in AI policy debates. Policymakers should be careful not to impose broad restrictions on the underlying tools when narrower conduct based rules can do the job.

It should preserve targeted safeguards for critical infrastructure. Where AI systems operate in especially sensitive contexts, lawmakers can require risk management, human override mechanisms, and similar precautions without undermining the broader right.

It should provide legal clarity. A state that protects computational freedom sends a signal to entrepreneurs, researchers, investors, and students that it is a place where innovation can proceed under clear rules rather than shifting administrative discretion. 

Why Explicit Protection Still Matters

Some critics will say we do not need a right to compute because the Constitution, property rights, and existing legal protections already do the work. But that objection misses the point. Rights that may seem implicit often need to be stated clearly when a new technology becomes a target of regulation.

While one could argue our right to compute is already implied, the current wave of AI regulation makes explicit protection necessary. Writing the principle into law gives courts, agencies, and future lawmakers a clear baseline that lawful computing is a protected activity, not a privilege to be narrowed whenever safety or convenience is invoked in the abstract. An explicit right to compute helps prevent regulatory creep by making clear that restrictions must be narrow, justified, and truly exceptional.

Without an explicit right to compute, states are considering licensing requirements tied to computing thresholds that could make it illegal for universities, startups, or independent researchers to train advanced models without government approval. 

The Bottom Line

As states consider how to govern powerful computing technology like AI, the right to compute offers a clear starting principle: People should be free to use lawful computational tools unless the government can show a strong and specific reason to restrict them.

That does not mean no rules. It means the burden should rest where it belongs, on the government, not on every builder, researcher, student, or business that wants to use modern tools to solve real problems.

In that sense, the right to compute is a digital age statement of economic liberty, technological openness, and lawful innovation.

Taylor Barkley is Director of Federal Government Affairs at the Abundance Institute, where he focuses on federal and state technology policy at the intersection of culture, technology, and innovation. Prior to joining the Abundance Institute, he served as Director of Technology and Innovation at Utah State University’s Center for Growth and Opportunity, Stand Together Trust, and held roles at the Competitive Enterprise Institute and the Mercatus Center. He is a graduate of Taylor University in Indiana and lives in the Washington, DC area.

Bryce Chinault is Director of State Government Affairs at the Abundance Institute, where he works with policymakers across the country to create space for emerging technologies to grow and reach their full potential. Before joining the Abundance Institute, he served as Director of External Affairs at the Yankee Institute for Public Policy and held roles at the George Washington University Regulatory Studies Center and the Mercatus Center. He earned a Master of Public Policy degree from George Mason University and is based in Newtown, CT.