Delaware’s draft bill would shield the owners of AI-run companies from liability for harm their AI causes, and the committee’s chairman wants to know whether Delaware weighed the risks and who is driving the plan.
Rep. James Comer (R-Ky.), chairman of the House Oversight Committee, is investigating a Delaware plan to create a new kind of company that an AI agent would run in place of people. Under the plan, the company’s owner generally couldn’t be held responsible for harm the AI causes.
Comer asked Delaware Gov. Matt Meyer for the state’s communications with outsiders who helped write the plan, including AI companies, and any records of how it assessed the plan’s safety risks. In his Oct. 26 letter, he said the plan “could be used to protect powerful AI companies and their investors from responsibility for reckless harmful actions.” The letter is a request, not a subpoena.
How the Delaware plan would work
A Delaware state committee released the plan as a draft bill in June. A single owner, either a person or a Delaware company, could set up what the bill calls an “Artificial Intelligence Company” and put an AI agent in charge of running it. Like any company, it could sign contracts, own property, and sue or be sued.
Someone harmed by the AI could sue the company. The owner would be on the hook only if they underfunded the company, used it for fraud or to knowingly break the law, or let it stray outside the business it was set up for.
Delaware would allow these companies only as an experiment. An owner would have to apply to a state official for permission to form one. Each approved company could do business for up to two years, and the official could order it dissolved for breaking the program’s rules. The experiment would end after 30 months, and lawmakers would then decide whether to keep this kind of company. Spotlight Delaware, a local news outlet, reported that the bill is likely to go to Delaware’s legislature in 2027.
Comer’s concern: AI companies could escape responsibility for harm under a law they’re helping write
Comer worries that an AI agent running one of these companies could harm business partners, consumers, or the public, and its owner would not have to answer for it. He says that protection would benefit powerful AI companies and their investors, and AI companies are helping write the plan.
Norm Ai, a company that builds AI agents, says it leads the work as Delaware’s “development partner.” Its chief executive, John Nay, wrote about the plan in a July op-ed in Fortune with Delaware’s Secretary of State, Charuni Patibanda-Sanchez. The two invited other companies that build AI agents to help.
Comer says Delaware hasn’t been open about the industry’s role. In September, Patibanda-Sanchez’s office declined to tell WHYY, a Philadelphia public broadcaster, about its working relationship with an AI company that contributed to the plan.
He also says there’s no sign Delaware looked at the dangers. The plan began in July 2025, when Meyer signed a resolution telling the state’s AI Commission to design a trial program for AI agents, which grew into this plan. In December 2025, the commission reported to Meyer that a subcommittee was drafting the trial’s rules. Comer says the report mentioned no work to study whether the plan was safe or what harm it could cause, and no advice from independent experts.
Comer’s probe could lead to federal limits on state AI laws
Comer asked for these records, covering January 2025 to now, by Nov. 9:
- the state’s communications with people and companies outside government about the plan, including its internal AI chat logs
- agreements with outsiders who helped write the plan
- any discussion of safety concerns, or of limiting legal liability for AI
- every draft of the plan and every comment on it
Comer said what he learns could shape federal bills, including ones that would limit what states can do on AI. A spokesperson for Meyer told Bloomberg Law the governor’s office is reviewing the letter.
