Delaware wants companies with no human at the helm

The state that charters about two-thirds of the Fortune 500 has drafted a company type run by software, with a human owner shielded from its mistakes. A House committee chair wants the paperwork.

*The state that charters about two-thirds of the Fortune 500 has drafted a company type run by software, with a human owner shielded from its mistakes. A House committee chair wants the paperwork.*

One sentence in a Delaware draft bill drew a House committee chair's attention. Per Ars Technica, the bill says: "The member is not liable for any act or omission that the AIC takes or fails to take, or that another person takes or fails to take at the direction of the AIC."

The AIC is an Artificial Intelligence Company, a new kind of business the bill would create, with an AI agent "designated to manage the business and affairs, and to conduct the business, of the AIC." The member is the person or company that funds it and collects the profits. Delaware's draft moves the answer to "who is responsible?" off humans and onto a legal entity run by software. That design choice is the center of Comer's letter, alongside a second worry he raised about regulatory capture.

Short answer: the AI-run company answers for harm, and its owner mostly does not

An AIC is a separate legal entity whose day-to-day affairs are managed by an AI agent instead of a person. It can sue and be sued in its own name. Take a hypothetical: an AIC signs a supplier contract and then fails to pay. The supplier's claim goes against the AIC. The funder behind it is generally not liable unless an exception applies, so the loss lands on whatever capital the AIC holds. That is what the Fortune op-ed means by "capital tied to liability."

The shield has limits the op-ed spells out. It "exists only inside the sandbox, and only for participants that follow the rules," and each AIC must tell counterparties it is "an authorized test entity." Banking is excluded. The sandbox is a 30-month supervised test, after which lawmakers decide whether to keep the program. Nothing here is law yet.

Supporters say a legal wrapper gives courts something to grab

Delaware Secretary of State Charuni Patibanda-Sanchez made the case in a July 14 Fortune op-ed co-written with John Nay, CEO of Norm Ai, a company that says it makes legal AI agents and leads the industry side of the plan.

Their starting point is that autonomous software will run companies whatever states do. "The answer is to wrap AI in legal form," they wrote. A legal identity creates "a defined target to which responsibility and damages can attach." Without a careful American home, they warn, "the activity will migrate offshore and onto anonymous infrastructure beyond the reach of any court."

Their guardrails: officials can suspend an AIC or ask the Court of Chancery to dissolve it, and consumer-protection and criminal law apply in full.

Critics say the wrapper removes the humans who keep a company in check

Both sides want accountability. They disagree on whether an entity that can be sued is enough when no person sits behind it. The supporters' tool, a defined target for damages, is the one the critics say leaves a gap.

York University law professor Martin Petrin put the objection in one line, per PYMNTS's summary of a Route Fifty report: "The main issue is really accountability because, in regular corporations, we have humans." Human directors and officers can be bound by fiduciary duties, which are legal duties to put the company's interests first. They can face personal liability and reputational damage. An autonomous AI has none of those incentives.

Sergio Alberto Gramitto Ricci, an associate law professor at Hofstra, pointed at what the draft leaves open, including how an AI-controlled company would relate to human managers. "How would managers report to the AI?" he asked. "These are aspects of governance that really make a corporation work."

Delaware has money riding on the answer. More than 2 million companies are incorporated there, generating roughly $2 billion in taxes and fees. Patibanda-Sanchez has said a new entity form would be "a huge benefit to our state's revenues." Ricci warned that if AICs cause significant harm, Delaware "might lose its gravitas as the state that moves in a conscious way."

Comer read the owner shield as a way to dodge blame for the software

Comer, who chairs the House Committee on Oversight and Government Reform, wrote to Gov. Matt Meyer in a letter that Ars Technica reported on Oct. 27. The committee "seeks assistance obtaining documents related to your state administration's efforts to establish a new form of corporate entity to be known as an Artificial Intelligence Company (AIC)."

Citing Patibanda-Sanchez's announcement, Comer wrote that an AI-run company would give "the human or entity capitalizing the AIC" "liability protections from potential harms to business partners, consumers, and the public, perpetrated by the controlling AI." The proposal, he said, "could be used to protect powerful AI companies and their investors from responsibility for reckless harmful actions, while also protecting non-transparent forces seeking anticompetitive regulatory capture behind state AI initiatives."

The letter also points past Delaware. Comer said the probe would "inform possible federal legislative efforts including proposals to appropriately limit state AI regulatory initiatives." That signals a possible preemption fight, where a federal law overrides what a state may do. Grit Daily notes the letter did not announce a bill or establish that the proposal violates existing law.

Delaware lawmakers see the bill in January, and Comer's request is still open

The congressional letter lands after more than a year of state work, starting with a July 2025 sandbox resolution. The collision comes down to who answers when the software gets it wrong. Delaware's draft puts that on the AIC itself, and Comer wants to know why a human owner should be shielded for it. Here is where each piece stands:

• Delaware's legislature. The proposal is expected to come before lawmakers after they reconvene in January, per PYMNTS. Until then it is a draft, and the critics' open questions, such as how much capital counts as adequate.
• Comer's request. The document demand is outstanding. A spokesperson for Meyer told Ars the office is "in the process of reviewing it." A drafted federal bill would turn a probe into a preemption fight, and none has been announced.
• The 30-month sandbox. It is meant to give the legislature "a full record for deciding how to legislate," per the op-ed.

Sources: Fortune op-ed, Jul 14 (the proposal's argument, the shield's limits); Ars Technica, Oct 27 (liability clause, Comer's letter, Meyer's office); PYMNTS, Sep 25 (the critique, January timing); Grit Daily, Oct 29 (what the letter did not do).