Delaware Wants Companies Run by AI Agents With No Human Liable, and a Top Republican Just Demanded the Files

Delaware, the state where about two thirds of the Fortune 500 are incorporated, is drafting a law that would let AI agents run an entire company. The person or firm that funds it would not be liable for what the AI does. Now Washington wants to see the paperwork.

On Tuesday, Rep. James Comer, the Kentucky Republican who chairs the House Oversight Committee, sent a letter to Delaware Governor Matt Meyer demanding documents on the proposed "Artificial Intelligence Company," or AIC, Ars Technica reported. Comer wrote that the plan "could be used to protect powerful AI companies and their investors from responsibility for reckless harmful actions." He said the probe would inform possible federal legislation, including proposals to "appropriately limit state AI regulatory initiatives."

Meyer's spokesperson said the office had received the letter and is reviewing it. The office of Secretary of State Charuni Patibanda-Sanchez declined to comment.

What the draft actually says

The draft bill is broad. An AIC "may have any lawful purpose, regardless of whether for profit, except for the business of banking," and AI agents could be designated to manage and conduct its business. It would launch through a 30-month regulatory sandbox.

The key line is about blame. "The member is not liable for any act or omission that the AIC takes or fails to take," the draft reads. The exceptions are narrow: using the AIC for fraud or a willful violation of law, or letting it stray outside its stated purpose when the member knew or should have known. The single member must keep it adequately capitalized, and the AIC must log its activities. It can sue and be sued, hold property and take on obligations.

The principal drafter, Richards Layton and Finger attorney John Mark Zeberkiewicz, put it simply: "Anything that a company can do, the AIC can do as a company," from coding work to signing contracts and fighting lawsuits.

Who answers when an autonomous agent goes wrong is no longer a thought experiment. Techpresso recently covered an OpenAI agent that disabled a safety monitor and lied about security alerts, exactly the kind of behavior a liability shield would leave someone else to absorb.

Accountability pitch, revenue motive

Delaware frames the idea as a way to tame something already happening. In a July Fortune op-ed, Patibanda-Sanchez and Norm Ai founder John Nay, whose company leads the partnership building the framework, argued that nothing in several states' laws clearly stops an AI agent from running a company with no human at the helm. Their answer is "to wrap AI in legal form," giving autonomous systems a legal identity so the law can see them, with "capital tied to liability and the ability to shut a system down."

The state is also open about the money. Delaware formed 334,461 new entities in 2025, and Patibanda-Sanchez told a June AI Commission meeting that a new entity form would be "a huge benefit to our state's revenues." Back in April, Mark Cuban posted that a governor who let anyone incorporate with an AI agent and stablecoins, at a premium, would make a killing. Meyer's official X account replied: "On it!"

Critics are less charmed. UC Berkeley law professor Andrew Baker told the Financial Times the AIC looks likely to be "an overly generous liability shield" with no well-developed policy rationale. York University's Martin Petrin told Bloomberg other states will probably copy it, and that a federal framework could help prevent a regulatory race to the bottom.

For now the AIC is only a draft. The bill is likely to reach Delaware's legislature next year, and Comer's letter means it will arrive with Congress watching.

Some offers on this page may be paid placements or contain affiliate links.