"No human at the helm": Delaware plan would let AI agents run entire businesses

US Rep. James Comer (R-Ky.), chairman of the House Committee on Oversight and Government Reform, sent a letter yesterday to Delaware Governor Matt Meyer requesting documents about the state's effort to create a new corporate form called an Artificial Intelligence Company, or AIC. Comer argued the proposal could let the human owners of AI-run firms escape responsibility for reckless or harmful actions taken by their agents, and said the committee's inquiry would inform "possible federal legislative efforts including proposals to appropriately limit state AI regulatory initiatives." A spokesperson for Meyer told Ars that the governor's office has received the letter and is reviewing it; a spokesperson for Delaware Secretary of State Charuni Patibanda-Sanchez declined to comment.
The AIC concept is set out in a draft bill that may go to the Delaware legislature, and it is unusual in scope. Under the draft, an AI agent could be "designated to manage the business and affairs, and to conduct the business, of the AIC," and the entity could pursue any lawful purpose except banking. The proposal would sit inside a "regulatory sandbox" lasting 30 months, giving legislators a test period before deciding on permanent rules. Delaware officials also frame the plan as a revenue opportunity: the state collects fees and taxes from entity formations and is already a leading home for corporate registrations.
Patibanda-Sanchez publicly announced the idea in a July 2026 Fortune op-ed co-written with John Nay, founder and CEO of Norm Ai, a company that builds legal AI agents for businesses. The op-ed argued that existing state law does not clearly stop an AI agent from running a company without a human at the helm, and that prohibition would be futile. Its stated approach is to give autonomous software a recognizable legal identity so that responsibility and damages have a defined target and the agent's conduct becomes traceable. The op-ed also described the liability structure in some detail: the AIC would be a separate legal entity managed day to day by an AI agent; it could sue and be sued in its own name, hold and dispose of property, and incur obligations. A single member, either a person or an entity, would be responsible for keeping the AIC adequately capitalized, and the AIC would have to keep a log of its activities. That member would be shielded from the AIC's debts except where it fails to capitalize the company or uses it for fraud or a willful violation of law.
The draft bill's liability language goes further. It says the member is not liable for any act or omission the AIC takes or fails to take, or that another person takes at the AIC's direction, with exceptions for fraud, willful violations of law, and cases where the member caused or allowed the AIC to do business that the member knew or should have known fell outside the AIC's statement of purpose. Comer's letter seizes on exactly this design, citing Patibanda-Sanchez's announcement as offering the entity funding the AIC liability protection from harms to business partners, consumers and the public caused by the controlling AI.
John Mark Zeberkiewicz, a corporate attorney at Wilmington firm Richards Layton and Finger who is the principal drafter, told Spotlight Delaware that the measure could let AI agents do nearly any business activity, from coding services to signing contracts or filing and defending lawsuits, and that "anything that a company can do, the AIC can do as a company." Spotlight Delaware reported the bill is likely to be introduced in the state legislature next year. Patibanda-Sanchez told a Delaware AI Commission meeting in June 2026 that new entity formations would be a significant boost to state revenues.
The proposal has drawn criticism from legal academics. Andrew Baker, a law professor at UC Berkeley, told the Financial Times the plan looks like an overly generous liability shield that seems unwarranted without a policy rationale that has been developed or justified well. Martin Petrin of York University in Toronto told Bloomberg that other states will likely follow with similar proposals to compete, and that a federal framework could help address safety concerns and a regulatory race to the bottom.
The political backdrop is tangled. Governor Meyer, a Democrat, publicly embraced the idea in a response to an April 2026 social media post by Mark Cuban suggesting states let people incorporate using an AI agent and stablecoins for a premium fee; Meyer's account replied "On it!" and linked to his July 2025 announcement of a broader AI sandbox covering corporate governance, biotechnology, healthcare, chemicals and financial services. In July 2025 Meyer signed a legislative resolution creating a sandbox framework for agentic AI, which led the Delaware Artificial Intelligence Commission to form a Sandbox Subcommittee that, per its December 2025 annual report, was working on potential draft legislation and had been briefed by Patibanda-Sanchez on her group's work. Comer's letter, in turn, signals possible federal legislation to constrain state AI regulation, which would test how far Congress can go in limiting state corporate and commercial law.
The stakes are amplified by Delaware's role in US corporate law: about two-thirds of Fortune 500 companies are incorporated there, and Patibanda-Sanchez says 334,461 new entities were formed in the state in 2025. Delaware's Court of Chancery and its body of case law are widely treated as the default rulebook for large companies, so a new entity type with an AI in the management seat would land in a jurisdiction whose precedents other states and companies watch closely. In practice, the draft's capital and logging requirements look like an attempt to keep a human anchor for solvency and an audit trail, while the liability carve-outs are narrow enough that they may not settle the harder questions, such as how a claimant proves an AI acted outside a statement of purpose, or who is accountable when an agent's transactions cascade. This suggests the sandbox's value to policymakers will depend on whether those questions get tested and recorded during the 30 months.
Why it matters: If Delaware adopts the AIC form, developers building autonomous agents that contract, pay suppliers and transact could for the first time operate inside a recognized legal wrapper, with a capitalized member behind it, rather than relying on human signatories. For product teams, the draft signals that agent design choices such as activity logs, defined purpose statements and shutdown controls may become compliance requirements rather than features. Comer's request and his hint at federal limits mean the timeline is uncertain, and companies weighing where to incorporate should expect the liability-shield question to be contested in both Dover and Washington.
Based on reporting from the original publisher. Visit the source for full context and later updates.
Publisher excerpt
Owners of AI-run firms would evade liability for "reckless" AI, US Rep. Comer says.