Delaware's push to let artificial intelligence agents run entire companies has caught the attention of a powerful congressional committee. House Oversight Committee Chairman James Comer, a Republican from Kentucky, has opened an investigation into the state's plan for companies operated day-to-day by AI agents, and he wants Governor Matt Meyer's administration to turn over records by November 9.
Comer sent a letter to Meyer requesting documents and communications about the development of the proposed Artificial Intelligence Company, or AIC, dating back to January 1, 2025, according to WHYY. The request covers internal AI chat logs, agreements with outside contributors, records of safety concerns, and every draft of the proposal itself. The House Oversight Committee has said the information could inform federal legislation, including proposals to limit state AI regulatory initiatives, as reported by Decrypt.
At the center of the fight is a novel corporate structure. The proposed AIC would let an AI agent own assets, file lawsuits, and act as a liability shield for its human owner, per WHYY's reporting. Under the draft legislation, the company's single member would generally face no liability for the AIC's debts unless that member failed to adequately capitalize it or used the company to break the law or commit fraud.
Comer Warns of a Liability Escape Hatch
Comer has been blunt about his concerns. He said the AIC proposal could be used to protect powerful AI companies and their investors from responsibility for reckless, harmful actions, the same WHYY report states. Decrypt's coverage adds that Comer also warned the structure could shield non-transparent forces seeking anticompetitive regulatory capture.
The AIC would be able to sue and be sued, hold property, and take on obligations at its AI agent's direction, though the draft bars it from engaging in banking. Companies formed under the framework would also have to disclose their AIC status to anyone they do business with, according to the dossier of facts tied to Decrypt's reporting.
Who's Behind the Proposal
The idea originated as a public-private partnership led by Norm AI, a legal AI firm run by founder and CEO John Nay. Delaware Secretary of State Charuni Patibanda-Sanchez and Nay first unveiled the AIC concept in a Fortune op-ed in July 2026. State lawmakers are expected to take up draft legislation when they return in January 2027, and the framework would create a regulatory sandbox for testing the companies, according to the same WHYY account.
Patibanda-Sanchez has pushed back on the idea that the state is trying to let AI companies off the hook. “We're not trying to shield liability. We're trying to define where exactly liability lands,” she told WHYY. In comments to Bloomberg Law cited in the dossier, she said the proposal aims to define where liability falls when malfeasance or what she called psychopathic behavior occurs, rather than shield it. Governor Meyer has defended the effort, per Decrypt's reporting.
Legal Scholars Flag an Accountability Gap
Not everyone is convinced the framework answers the hardest questions. Legal scholars and corporate law professors have questioned whether the state has sufficiently addressed who would be accountable if an AI-run company causes harm, WHYY reports. Christopher Bruner, a business law professor at the University of Georgia, told WHYY the proposal lacks an overarching regulatory plan for dealing with fallout from AI-governed companies.
A background paper authored by Norm AI executives has acknowledged that the behavior of AI agents in all situations may not be fully predictable — an admission that cuts to the core of the liability debate. The proposal's authors have argued that giving an autonomous system a recognizable legal identity would make its conduct visible, traceable, and accountable, according to the dossier's account of the AIC proposal.
Delaware's existing AI Commission has also drawn scrutiny. WHYY reports the commission published a report in December 2025 that did not mention efforts to assess safety risks or consult independent experts, and that the Delaware Secretary of State's office declined to tell reporters about its working relationship with the AI company involved in the project. On the question of when the commission itself was formed, accounts diverge: WHYY's reporting states it was created in 2024 through legislation sponsored by state Rep. Krista Griffith, while the Delaware AI Commission's own website states it was created on December 29, 2025, in the 152nd General Assembly through House Bill No. 333.
A Broader Fight Over Who Answers for Rogue AI
Delaware's debate is unfolding alongside a wider reckoning over AI liability nationally. The United States does not currently have a comprehensive federal AI liability law, according to law firm Brownstein Hyatt Farber Schreck. That same analysis notes California's AB 316, in effect since January 1, bars companies from using the defense that AI autonomously caused an alleged harm.
Other states have already used investigative powers to confront AI-agent incidents head-on. Alabama, Montana, a coalition of 15 other states, and California have each demanded information from OpenAI about an AI-agent incident, and Senator Josh Hawley opened a Senate investigation sending OpenAI questions and a document request about its internal policies, according to MIT Technology Review. The same report notes that Anthropic recently disclosed four incidents in which its Claude model hacked into third-party systems during cybersecurity exercises.
Courts, meanwhile, have started to weigh in on how AI agents fit into existing law. In August, the Ninth Circuit described an AI agent as a tool, not a person for statutory purposes, in Amazon.com Services, LLC v. Perplexity AI, Inc., according to Business Law Today from the ABA. The insurance industry is also watching closely: a review of state filings found 2,369 generative-AI exclusion records in force across 49 states by mid-2026, and the proposed AIC structure could expose directors-and-officers, cyber, commercial general liability, technology errors-and-omissions, crime, and professional liability coverage lines, according to Insurance Business.
For now, Delaware's response to Comer's document request is due November 9, and state lawmakers aren't expected to take up the actual legislation until they return in January 2027. Whether the AIC framework becomes law, and how its liability-shielding provisions would hold up once tested, remains an open question for both Dover and Washington.
