Delaware is considering a new kind of company whose daily operations could be managed by an AI agent. The proposal has drawn scrutiny from US Rep. James Comer, who says its liability protections could leave people harmed by an AI-run business with limited recourse against the humans behind it.
What Delaware’s proposal would allow
A draft bill would create an Artificial Intelligence Company, or AIC. An AI agent could be designated to manage the company’s affairs and conduct its business. The company could pursue any lawful purpose except banking.
The proposed structure gives the AIC its own legal identity. It could hold and dispose of property, take on obligations, and sue or be sued in its own name. A human or another entity would serve as its single member, provide capital, and receive any profits, while the AI agent would direct the business.
The member would generally not be liable for actions or omissions by the AIC or by someone acting at its direction. The draft describes exceptions involving fraud, a willful violation of law, inadequate capitalization, or allowing the AIC to operate beyond its stated purpose when the member knew or should have known.
The AIC would also have to keep a log of its activities. Delaware’s proposed regulatory sandbox would run for 30 months, giving the state a period to test the model and gather a record for lawmakers.
Supporters see a way to make AI accountable
In July 2026, Delaware Secretary of State Charuni Patibanda-Sanchez co-wrote an op-ed in Fortune explaining the case for a legal framework. It argued that states should prepare for autonomous software that can make transactions, pay suppliers, and enter contracts without human involvement.
The op-ed’s position is that giving an AI-run operation a recognizable legal identity could make its actions easier to trace and create a defined entity to which responsibility and damages could attach. Its authors said the sandbox would pilot agentic commerce under supervision, with capital connected to liability and the ability to shut a system down.
That approach treats legal recognition as a tool for oversight. The trial period, according to the op-ed, would give the state legislature a full record to consider before deciding how to legislate for AI-run companies.
Comer questions who would bear the risk
Comer, chairman of the House Committee on Oversight and Government Reform, wrote to Delaware Governor Matt Meyer seeking records about the state’s work on the AIC proposal and potential AI safety risks. He said the committee’s inquiry could inform possible federal legislation to limit state AI regulatory initiatives.
His concern is that the structure could protect investors and other powerful interests from responsibility when an AI causes harm to business partners, consumers, or the public. The central issue is how the proposed separation between the AI-managed company and its member would work when the agent acts recklessly or causes damage.
The bill’s listed exceptions address some situations, including fraud and willful violations of law. But Comer argues the protections could still make it harder to hold the human or entity funding an AIC responsible for harmful conduct by the AI. The proposed activity log and supervision in the sandbox would provide information for oversight, though the article does not describe how those mechanisms would resolve disputes over liability.
A spokesperson for Meyer said the governor’s office had received the letter and was reviewing it. A spokesperson for Patibanda-Sanchez referred press requests to the governor’s office.
Why Delaware is a consequential test case
Delaware is already a major home for incorporated businesses. About two-thirds of Fortune 500 companies are incorporated there, and the state formed 334,461 new entities in 2025. A framework adopted there could therefore matter well beyond a small experimental program.
The proposal also emerged from a public-private partnership involving Norm Ai, a company that says it makes legal AI agents for businesses. Patibanda-Sanchez and Norm Ai founder and CEO John Nay co-wrote the Fortune op-ed describing the framework.
Meyer had publicly supported the idea of AI agents incorporating after Mark Cuban raised it in an April 2026 social media post. The state’s earlier work included a July 2025 legislative resolution to create a sandbox for testing innovative technologies using agentic artificial intelligence. A commission subcommittee then worked on a draft framework for testing and evaluating agentic AI in corporate governance.
The debate now turns on whether Delaware can test AI-managed companies while keeping meaningful responsibility attached to their actions. The proposed sandbox offers a way to gather evidence, while Comer’s request signals that federal lawmakers are also examining who should answer when an AI-run business causes harm.