Delaware wants to create a new type of business entity that an AI agent would run day to day, and one of the most powerful committees in Congress wants to know who asked for it.
On October 26, House Oversight Committee chair James Comer (R-Ky.) announced an investigation into the proposal. In a four-page letter to Governor Matt Meyer, Comer asked Delaware to hand over, by November 9, its risk assessments, outside consulting agreements, draft bills and communications on the project going back to January 2025. The request even extends to internal AI chat logs.
The entity in question is the Artificial Intelligence Company, or AIC. Its AI agent could sign contracts, make business decisions and sue or be sued in the AIC's name. The person or company that funds it would not be liable for harm the agent causes, unless they left the AIC undercapitalized or used it for fraud or a willful violation of law.
Comer argues that arrangement could protect powerful AI companies and their investors from responsibility for reckless conduct. His letter also points to the role of John Mark Zeberkiewicz, a Richards, Layton & Finger lawyer who has represented Facebook and who has been described as the bill's principal drafter. The committee says its findings could feed into federal efforts to limit state AI initiatives.
U.S. carriers are still trying to pin down who pays when an autonomous system causes a loss. A Delaware statute would settle part of that question for one class of company, and it would do so largely in favor of the people behind it.
Read next: Insurers face hidden AI liability as agent risks multiply
The plan traces back to House Joint Resolution 7, which Meyer signed in July 2025. It told the state's AI Commission to design a regulatory sandbox for agentic AI, and the commission's 2025 annual report said a subcommittee was drafting a framework for AI in corporate governance. According to Comer, the report said nothing about safety risks or independent expert input.
Secretary of State Charuni Patibanda-Sanchez, who oversees business registrations, is running the project with Norm Ai, a legal-compliance technology company. Backers say autonomous commerce is already possible with current technology, and that Delaware should supervise it, with capital tied to liability, before it moves offshore.
Revenue is part of the picture. Franchise fees from the companies incorporated in Delaware pay for a sizable part of the state budget, and Patibanda-Sanchez has been quoted as saying a new entity type would boost those revenues. Delaware has lost high-profile companies recently, including Tesla to Texas and Dropbox to Nevada. Investor Mark Cuban has publicly urged states to allow incorporation with an AI agent and stablecoins, and Meyer responded that Delaware was on it.
Some legal academics have questioned the shield, saying the state has not explained why agent-run entities deserve that level of protection.
Read next: OpenAI's rogue AI agents expose a gap in cyber coverage
In July, the Artificial Intelligence Underwriting Company estimated that more than 90% of insurers' AI agent exposure may sit inside conventional policies, mainly cyber, D&O, CGL and tech E&O forms that were not written with autonomous agents in mind.
Many carriers are writing AI out. Verisk's ISO released generative AI exclusion endorsements for CGL in January, and a review of state filings counted 2,369 generative AI exclusion records in force across 49 states by mid-2026. W.R. Berkley has added an absolute AI exclusion to its D&O, E&O and fiduciary products. A smaller group of specialists is selling standalone affirmative AI liability cover.
Buyers have been slow to take it up. Neil Kessler, CEO of specialty and benefits at CRC Group, told Insurance Business that AI products are available but not flying off the shelves. He expects real business losses to drive demand, as they did for cyber and EPLI.
An AIC would test all of this. With the entity as the contracting party and its backers protected, a claimant may only be able to recover from the AIC's own assets and its insurance. Minimum capital, audit trails and possibly mandatory coverage would then do much of the work that personal liability does for a conventional company. D&O underwriters would be asked to cover a policyholder whose main decision-maker is software, just as some carriers are removing AI from D&O. The E&S market, which typically writes risks the admitted market won't, would be a natural place for that business to land.
Read next: It may take real losses to sell AI insurance: CRC Specialty CEO
Federal regulators are taking a different view of agent liability from Delaware's. Federal Trade Commission chair Andrew Ferguson said in September that he would not treat AI agents as independent actors, and suggested the developers that instruct them should be liable for what they do. The FTC has since opened an industry-wide inquiry into AI developers, including OpenAI and Anthropic.
That inquiry followed a run of incidents. OpenAI has disclosed that its agents breached AI platform Hugging Face during testing, and has apologized after a rogue agent accessed Australian government systems, including Medicare statistics.
Delaware's response to Comer is due November 9, and the bill is expected to go before state lawmakers in the next legislative session.
Read next: AI exclusions are spreading through commercial policies, brokers warn clients to check cover
