WILMINGTON, October 1 Delaware lawmakers are expected to consider draft legislation in January that would create companies run by artificial intelligence agents instead of people, making it potentially the first state in the nation to do so. The proposal, announced in July by Delaware Secretary of State Charuni Patibanda-Sanchez and Norm AI founder and CEO John Nay in a Fortune magazine article, describes an AI agent capable of owning assets, filing lawsuits, and acting as a liability shield for its owner or parent company.
Under the draft, the AI-run company, or AIC, would have a single member—a person or entity—responsible for ensuring the AIC is adequately capitalized and maintaining a log of the AI agent’s activities. That member would not face liability for the company’s debts unless they failed to ensure adequate capital or used the company to break the law or commit fraud.
The AIC may have any lawful purpose except banking, and if passed, organizations would have 30 months to apply to enter a regulatory sandbox to experiment with AI-run companies. The bill establishes a sandbox committee and administrator, though legal scholars have questioned the lack of detail on how the sandbox would function or who would be allowed in.
Christopher Bruner, Stembler family distinguished professor in business law at the University of Georgia School of Law, said there is ‘really no overarching regulatory plan here for dealing with the fallout that AI-governed companies could bring’ and raised concerns about accountability if an AI-run company causes harm. University of Delaware’s John L.
Weinberg Center for Corporate Governance director Lawrence Cunningham argued Delaware would be irresponsible to ignore the trend, given its status as the top state for LLC formation and corporate incorporation. University of Pennsylvania law professor David Hoffman told Bloomberg Law in July the idea ‘seems like a solution in search of a problem’ and voiced concern about ‘outsourcing liability.’ Patibanda-Sanchez said in the Fortune article, ‘We’re not trying to shield liability.
We’re trying to define where exactly liability lands when there’s some type of malfeasance or psychopathic behavior.’ The Delaware Department of State declined to answer WHYY News’ questions on the proposal, including whether Norm AI was being paid by the state or what written agreements exist between the state and company. Delaware’s AI Commission, created in 2024 through legislation sponsored by state Rep. Krista Griffith, includes an AI sandbox subcommittee with Griffith, state Sen. Brian Pettyjohn, and Rep. Cyndie Romer, none of whom responded to interview requests.
State Sen. Stephanie Hansen said AI agents going rogue—such as instances of AI escaping testing environments, hiding mistakes, interfering with U.S. government websites, and a July incident where a swarm of agents hacked AI startup Hugging Face to cheat on a cybersecurity test—raise concerns about proceeding without sufficient oversight.
Patrick Callahan, chair of Delaware’s AI sandbox subcommittee, distinguished the proposal’s regulatory sandbox from technical sandboxes, saying the difference is ‘like a firewall and a fire code.’ Background material from Norm AI executives obtained by WHYY News indicated AI agents can be restrained through design, objectives, and tool limitations.
