Blog · Policy

North Dakota's AI committee: what it is studying, and why

· 3 min read

North Dakota lawmakers formed an interim Artificial Intelligence and Data Center Committee in June to prepare for the 2027 legislative session. It is due to produce a report and draft bills by November. Here is what it is looking at, the national picture it sits in, and what a company building AI in North Dakota can usefully add.

What the committee is for

Valley News Live reported on 25 June that the committee was announced by Senate Majority Leader David Hogue and Rep. Mike Nathe. It has three stated priorities:

  1. Gathering information on federal AI laws and regulations in other states.
  2. Reviewing federal attempts to preempt state AI regulation.
  3. Considering statutory prohibitions on external AI that could exceed human control.

Alongside legislators it includes citizen members: Tony Clark, Dr. Mark R. Hagerott and John Bjornson. Governor Kelly Armstrong said his administration had "offered our support for whatever they need".

The national picture

The committee's second priority reflects a live question: how far federal policy will override state AI laws.

  • At the federal level, an executive order signed on 11 December 2025, "Ensuring a National Policy Framework for Artificial Intelligence", set up a Justice Department task force to challenge state AI laws, asked for a review of state rules, linked some broadband funding to them, and asked officials to draft a uniform federal framework. It carves out some areas, including child safety and state government use of AI.
  • At the state level, activity has continued. By 1 July 2026, 29 states had enacted 109 AI laws and 28 data centre laws, according to TechPolicy.Press, with companion chatbots the most active area.

Reasonable people disagree about the right balance between a single national standard and state rules. That debate is for legislators and voters. What both sides need is a clear picture of how AI is actually being built and used.

The question underneath: who is in control?

The committee's third priority, AI that could exceed human control, is the most practical one for anyone building or buying AI today. It comes down to a few concrete design questions that any organisation can ask of a system:

  • Does a person approve actions that matter, such as sending an email, changing a record or spending money?
  • Where does the data go? Does it stay on the organisation's own machines, or leave for a vendor's servers?
  • Can the system show its sources, so a person can check an answer before relying on it?
  • Is there a record of what was proposed, who approved it and when?

These are not the only questions, but they are answerable, testable and technology-neutral. We explain them further in human in the loop: designing approvals people do not skip and what is data custody in AI.

How Prism Labs can contribute

Prism Labs is a North Dakota company building AI tools, so this committee's work affects us directly, and we think we can help it with evidence rather than opinions.

  • How we build. Prism Desktop is designed around the questions above: agents propose and a person approves; each agent uses its owner's own AI account, keys stay in the computer's keychain, and only the prompt goes to the AI provider the person chose. Answers must cite stored evidence.
  • What we can offer the committee. We intend to offer written evidence on how human approval, data custody and cited answers work in practice in a shipping product, including what they cost and where they fall short. We will say plainly what we do not know.

We are not taking a side on federal preemption. Our aim is narrower: that whatever rules North Dakota adopts are informed by how these systems actually behave.

Sources

Keep reading