1. Comer opens an inquiry into the Delaware AI company proposal
Representative James Comer, the Kentucky Republican who chairs the House Oversight Committee, has opened an investigation into a Delaware proposal for a Delaware AI company: a new legal entity whose daily operations would be directed by an artificial intelligence agent. The committee announced the inquiry on Monday, October 26, 2026, and gave Governor Matt Meyer until November 9 to hand over a broad set of records.
Comer's letter to the governor centers on the proposed "Artificial Intelligence Company," abbreviated AIC. His central objection is that the structure would insulate the people and organizations standing behind it from responsibility if the AI agent running the business causes harm. The committee's public statement on the matter said the chairman was seeking documents, among them internal AI chat logs, in order to establish who is behind the plan.
The probe places a state-level experiment in corporate law in front of a congressional committee, and it does so at a time when Washington is already debating how far states should be allowed to go in regulating artificial intelligence. The committee says the material it collects could feed into federal legislation, including measures that would restrict state-level AI rules.
2. What is an Artificial Intelligence Company?
An Artificial Intelligence Company, as outlined by its authors, is a business entity that could be managed by an AI agent rather than by human directors making each decision. Such an entity could sue and be sued, own property, and enter into obligations at the direction of its AI agent. In other words, it would have the ordinary legal personality of a company, with software holding the steering role.
The design includes a human anchor. A single member would have to keep the entity adequately capitalized, and that member would be shielded from the company's debts. The shield is not absolute: it would not apply if the member left the company undercapitalized, or if the entity were used to commit fraud or a willful violation of the law. Limited liability of this kind is a familiar feature of corporate law, and the proposal borrows its logic while changing who, or what, makes the operating decisions.
The proposal also carries guardrails. AICs would be allowed to operate only inside a 30-month sandbox supervised by state officials together with outside experts. They would be required to disclose their status to anyone they deal with, and they would be barred from banking. A regulatory sandbox is a controlled environment in which a new type of business can run under supervision before any permanent rules are written, and this one is meant to give the state time to observe the entities before deciding what to do next.
3. Who proposed the structure and why?
The idea was introduced in July 2026 in an opinion essay written by Delaware Secretary of State Charuni Patibanda-Sanchez and John Nay, chief executive of the legal AI company Norm Ai. The essay described the effort as a public-private partnership led by Norm Ai, which means a private firm is deeply involved in shaping a framework that a state government may eventually enact.
The authors said the purpose of the proposal is to "wrap AI in legal form." Their argument is that giving an autonomous system a recognizable legal identity would make its conduct visible, traceable and accountable. Under that reasoning, an AI agent that acts through a registered entity, with a responsible member behind it and a duty to identify itself to counterparties, is easier to oversee than one operating through informal arrangements or through accounts that nobody can tie to a legal person.
Delaware's position in the debate carries extra weight. A large share of American public companies are incorporated in the state, and its corporate statutes and courts are closely watched across the business world. A new type of entity created there would be noticed well beyond its borders, which helps explain why a congressional chair has taken an interest.
4. Why does Comer say the plan could shield AI firms from liability?
Comer's letter argues that the AIC could protect powerful AI companies and their investors from being held accountable for reckless harm. In the letter he also raised concern about what he described as non-transparent forces pursuing regulatory capture through state-level AI initiatives. Regulatory capture refers to a situation in which the industry a rule is meant to govern ends up shaping the rule to its own advantage.
The chairman's worry rests on the way the structure assigns responsibility. If an AI agent is the decision-maker and a single capitalized member is shielded from the entity's debts, injured parties could find it difficult to reach the developers, deployers or financial backers who built and funded the system. Comer's letter treats that outcome as a risk of the design and not merely as a theoretical edge case.
The letter also questions the role of outside contributors. It notes that a December 2025 report from Delaware's AI Commission did not address how safety risks would be assessed or whether independent experts would be consulted. It adds that the Secretary of State's office declined to tell reporters about its working relationship with an AI company that is involved in the project. Together, those points frame the inquiry as a question about process and transparency as well as about liability.
5. What records is the committee requesting?
The document request is wide. Comer asked Governor Meyer for all records dating from January 2025 that relate to the development of the AIC concept. The list includes internal AI chat logs, agreements with outside contributors, records of safety concerns that were raised, and every draft of the proposal.
The inclusion of AI chat logs is notable. It suggests the committee wants to see whether artificial intelligence tools were used in drafting the proposal itself, and what role they played in shaping it. The request for contributor agreements points at the same question from another direction, asking who was involved and on what terms.
The deadline is November 9. Congressional oversight letters of this kind are usually the first step in a longer process, and they do not by themselves carry legal penalties. The response from the state, and the completeness of any records it provides, will help determine whether the committee pursues the matter further.
6. How Delaware officials are responding
Delaware officials reject the characterization that the entity is built to dodge accountability. Patibanda-Sanchez said in July that the plan is intended to define where liability falls when misconduct occurs, and she put it plainly: "We're not trying to shield liability." The state's view, in short, is that the framework answers a question that the law currently leaves open, which is who answers for the actions of an autonomous system.
Governor Meyer defended the effort in a September radio call-in appearance. He said companies run by artificial intelligence are coming and that ignoring the subject is not an answer. His comments describe the project as an attempt to get ahead of a development that is likely to arrive whether or not the law is ready for it.
Outside voices have been more cautious. Christopher Bruner, a business law professor at the University of Georgia, said in September that there is no overarching regulatory plan for the fallout from companies governed by AI. His remark does not reject the Delaware approach outright, but it points to a gap that the sandbox is meant to help fill and that critics say remains unresolved.
7. Why the liability question reaches beyond Delaware
Corporate law has always had to decide who answers when a business harms someone. Limited liability lets investors put money into a company without risking their personal assets, and courts have long allowed that protection to be set aside in cases of fraud or abuse of the corporate form. The Delaware proposal keeps those exceptions for the single member, but it adds a new variable: a decision-maker that is neither a person nor a traditional board.
That variable is what makes the dispute difficult to settle. Supporters can argue that a registered entity with a capitalized member and mandatory disclosure gives injured parties a defined target for a claim. Critics can reply that the same arrangement places a thin layer of capital between the victim and the firms that built the underlying model. Both readings rest on the same text, and the sandbox period is meant to show which one holds up in practice.
The inquiry also arrives while lawmakers in Washington weigh whether federal rules should limit what states may do about artificial intelligence. A state plan that critics describe as a liability shield offers advocates of federal preemption a concrete example to point to, while defenders of state experimentation can point to the sandbox, the disclosure duties and the ban on banking as evidence that Delaware is moving with care.
8. What happens next for Delaware AI companies
The next fixed date is November 9, when the state's documents are due to the committee. After that, the committee could review the material, request additional records, or move toward hearings, though none of those steps has been announced.
On the Delaware side, state lawmakers are expected to consider draft legislation when they return in January. If a bill advances, the AIC proposal would move from an idea described in essays and interviews to a concrete statute with defined capital requirements, disclosure duties and sandbox terms, and the details of those provisions would then be open to debate.
The outcome matters for anyone watching how autonomous software agents will fit into commercial law. Crypto and fintech observers have a particular interest, because AI agents that hold funds, sign contracts and transact on blockchains are an active area of experimentation, and the question of which legal entity stands behind such an agent is still open. Whether Delaware's model survives federal scrutiny may influence how other states, and Congress, approach the issue.