Delaware's plan for AI‑run companies draws a House investigation

The head of the US House Oversight Committee has opened an investigation into a Delaware plan that would let an AI agent run a company day to day. Under the proposal, the human or entity behind the company would get liability protections from harms caused by the controlling AI. Rep. James Comer, a Kentucky Republican, wants Delaware to hand over documents by November 9.

You asked whether this is a feature worth turning on. It isn't something you can switch on yet. Delaware's framework remains a proposal, not an enacted law. So this piece covers what the plan says, who is worried, and what comes next. I haven't tested anything here, because there is nothing to test.

What Comer said and asked for

Comer's concern is about accountability. In his letter he wrote that the plan could be used to protect powerful AI companies and their investors from responsibility for reckless harmful actions. He also raised the risk of "regulatory capture", where hidden interests use state AI rules to lock out competitors, according to WHYY.

The request is wide. The four-page letter to Gov. Matt Meyer asks for risk assessments, outside consulting agreements, draft bills and communications going back to January 2025, and even internal AI chat logs. Meyer's deputy press secretary told WHYY the office is reviewing the letter.

Comer also says Delaware's own paper trail is thin. He points to a December 2025 report from the state's AI commission. He says the two-page report did not mention any assessment of safety, harms or risks, or any consultation with independent experts. One outlet, Superpower Daily, adds a caveat: that omission is not proof that no assessments happened.

How the Delaware plan would work

The new entity is called an Artificial Intelligence Company, or AIC. Here is what the sources say about it:

• What the agent can do. Its AI agent could sign contracts, make business decisions and sue or be sued in the AIC's name.
• Who is liable. According to WHYY, the AIC would have a single member, a person or entity responsible for keeping it adequately capitalised. That member would face no liability for the company's debts unless it failed to do so or used the company to break the law or commit fraud.
• Record keeping. WHYY reports the AI agent must keep a log of its activities.
• Limits. The draft says an AIC may have any lawful purpose, with the business of banking excluded.
• Sandbox. WHYY says organisations would have 30 months to apply to enter a regulatory sandbox, overseen by a sandbox committee and administrator. Other write-ups describe the 30 months a little differently. The D&O Diary calls it a 30-month test period starting from passage of the legislation. Check the bill text once it is formally introduced.

One practical point for anyone hurt by an AIC. If the owner qualifies for liability protection, claimants may have to rely on the AIC's own assets and insurance for compensation.

Who is behind it

The idea has been public since the summer. WHYY reports that Secretary of State Charuni Patibanda-Sanchez and Norm AI founder and CEO John Nay announced the proposal, and a public-private partnership, in Fortune in July. The principal drafter is John Mark Zeberkiewicz, a lawyer at Richards, Layton & Finger. Spotlight Delaware quotes him on the scope: "Anything that a company can do, the AIC can do as a company."

Zeberkiewicz also described the appeal in plain terms. In the same report, he compared the shield to any limited liability company, which exists so owners aren't automatically liable for the entity's debts. The law firm told WHYY that practising attorneys have always helped develop Delaware's corporate statutes and that the General Assembly makes the final decisions.

Supporters say waiting is the riskier option. Spotlight Delaware reports that AI commission chair Patrick Callahan said agents are already transacting, citing Morgan Stanley giving agents access to stock trading platforms. He said no clear party is held accountable when things go wrong.

Critics and defenders

Legal scholars are split. WHYY quotes University of Georgia law professor Christopher Bruner saying there is "really no overarching regulatory plan" for the fallout from AI-governed companies. It also quotes University of Pennsylvania law professor David Hoffman, who told Bloomberg Law in July that the idea "seems like a solution in search of a problem."

On the other side, Lawrence Cunningham of the University of Delaware's Weinberg Center for Corporate Governance argued the state would be irresponsible to ignore the idea, given Delaware's place as the top spot to form companies.

Transparency is the other sore point. WHYY says Patibanda-Sanchez declined to answer its questions about the relationship between Norm AI and the state, and Nay did not respond to a request for comment. Comer cited that reporting in his letter.

What happens next

Delaware's response is due November 9. State lawmakers are expected to consider the bill next year, so any sandbox is months away at best. WHYY also notes that the letter could signal Congress is exploring limits on state AI regulatory initiatives.

For now, the takeaway is simple. No one can form an AIC today. The open question is whether lawmakers in Dover, and now Washington, think shielding the owners of AI-run companies is a sensible experiment or a loophole. I'll update this story when Delaware responds.