Montana Wants to Guarantee Right to Compute, but Is the Law Ready for Reality?

Key takeaways:

• Montana has made right to compute' a fundamental right, allowing residents to use AI systems without government interventions.
• Deployers will have to ensure a manual override on AI systems that run critical infrastructures.
• The law lacks several provisions such as oversight body and penal provisions, which can lead to a lot of litigation once it comes into effect.

Montana has become the first US State to officially pass a right to compute' law, known as the Montana Right to Compute Act (MRTCA).

MRTCA gives Montana residents the right to use computation resources' on their private property without undue government intervention. This includes activities like running blockchain nodes, AI modeling, and distributed storage systems.

Section 3 of the MRTCA says that Government actions that restrict the ability to privately own or make use of computational resources for lawful purposes must be limited to those demonstrably necessary and narrowly tailored to fulfill a compelling government interest."

Careful reading of the section highlights the fact that the burden of proof has been put on the law enforcement agencies if they want to restrict an individual's right to use computational resources. Any government intervention, if necessary, should be adequately argued by the authority.

Certain uses of such compute power have been classified as against compelling government interest,' including mechanisms that deceive or defraud the public, or activities involving the distribution of deepfakes and other harmful synthetic content that affect individuals, especially minors.

Senator Daniel Zolnikov, the proposer of MRTCA, was delighted on passing of the bill, calling it a key step in defending individual liberty. The Act will ensure that every Montanan can access and control the tools of the future.'

Mandatory Shutdown Mechanism: Keeping Human Control Over AI

An important provision of the Act is the inclusion of a shutdown mechanism.' Whenever critical infrastructure facilities are run by AI systems, the deployer must ensure the capability to disable the AI's control over such infrastructure and switch to human control within a reasonable amount of time.

The law also requires deployers of AI to implement, test, and annually review a risk management policy as per approved and latest AI Risk Management frameworks.

If you read between the lines, there are a few key implications of this provision. First, the lawmakers acknowledge and expect companies to use advanced AI systems for critical infrastructure such as electric substations, water treatment facilities, oil refineries, gas compressor stations, and so on.

At the same time, the Act accepts that AI systems are not perfect and are prone to system crashes or bugs, which is why it's important to incorporate human intervention in the entire process. If things go south, the deployer should ensure that human control is possible and can be regained until the AI systems are fixed.

The MRTCA's Biggest Problems: No Oversight, No Penalties, No Clarity

Our honest take of the MRTCA? In its current form it seems to be a half-baked piece of legislation that could do more harm than good. There are no provisions to ensure thorough compliance, and no guidelines outlining what the overseeing authority can and can't do. Most importantly, there are no penal provisions in the entire Act, which is yet another can of worms waiting to be opened.

Although the legislation provides a right, who exactly will protect it and who will have to be held accountable if this right is violated? Who's responsible for enforcing AI safety requirements on local compute? There are a lot of unanswered questions in MRTCA which could ultimately lead to several drawn-out litigations down the line.

Take the shutdown mechanism for example. If the deployer' ignores the shutdown mechanism requirement and an incident happens because of it, what are the repercussions? In this case, enforcement will have to happen through litigation instead of a clearly defined process, which means a lot more downtime until matters can be settled in courts.

To gain some perspective, one can look at what happened with the Arkansas Right to Mine Act. Passed in 2023, the Act aimed at protecting crypto miners and motivating them to set up mining rigs. In the process, the Act blocked local law enforcement from restricting resource usage such as electricity, water, and even noise pollution.

As a result of ARMA, many mining firms were set up in residential areas with unchecked levels of noise and resource consumption. The city authorities tried to intervene and regulate the situation, but the bill prevented them from doing so.As you'd expect, things went to court. Jones Digital, LLC, a crypto mining firm in DeWitt sued the county after local officials tried to pass an ordinance to restrict noise levels to 55dB in the day and 45dB in the night.

Jones Digital called this discriminatory towards mining firms and leaned on the Right to Mine Act. The company was ultimately allowed to resume operations and the county had to pay $90K as compensation. In the wake of similar cases, cities had to sue the state to regain their authority, and the Act was eventually rolled back within just a year.

Unfortunately, it seems that MRTCA has all the ingredients to cook up another Right to Mine' fiasco. Just like crypto mining, setting up high-compute AI systems locally requires huge amounts of electricity, generates excessive noise through cooling towers, and produces serious industrial heat from elaborate GPU racks.

And since the Act has no resolution mechanism, this could again lead to residents suing AI companies, while city authorities will be able to do little to help the situation, courtesy of MRTCA's lack of clarity.

The Supply Shortage and the Spirit of the MRTCA

Look, the MRTCA is undoubtedly a positive step in breaking AI free from the grip of capitalistic conglomerates. But, as things stand currently, the law just doesn't seem to be well thought out.

While local compute have been made a fundamental right, there are little to no practical means for residents to exercise this right.

For example, the core need of any AI system, even the most basic ones, is a powerful GPU, like the ones manufactured by Nvidia or TSMC, or rather rows and rows of these GPUs. And it's no hidden secret that the industry has been fighting hard for every last AI chip coming out of these foundries.

Nvidia's inventories have been sold out until 2026 with long wait lists, whereas TSMC's CEO has already stated that inventories will remain tight throughout 2025 and even into 2026.

So, when data centers and large players are struggling to secure powerful GPUs, we cannot realistically expect Montana entrepreneurs or innovators to get their hands on high-end AI chips without knowing the full implications of their investments.

What laws like MRTCA do is put more pressure on the supply chain, increasing demand and compute costs in the process. So, while the bill could attract more AI professionals to Montana, they'll likely be met with roadblocks sooner than they expect.

Plus, the cost of GPUs doesn't even remotely resemble consumer-level hardware costs. For instance, NVIDIA's H100 can cost anywhere between $25K and $40K, while the newer H200 can go all the way up to $65K. The kind of funds required for this level of massive investment simply isn't within reach for ordinary residents.

The all-important question, therefore, is who does this law actually benefit? Is it intended to empower local entrepreneurs and spark innovation, or is it designed to attract AI tech giants into Montana?

It seems quite improbable that the lawmakers are unaware of the high setup costs or shrinking supply chains. So, even though residents have been given a right' to participate in the AI race, that doesn't seem to be the spirit of the law.

The post Montana Wants to Guarantee Right to Compute, but Is the Law Ready for Reality? appeared first on Techreport.