THE ESSENTIALS IN 30 SECONDS

  • The draft would create an 'Artificial Intelligence Company' (AIC) whose operations are run by an AI agent, tested in a 30-month regulatory sandbox.
  • The AIC's single member would generally be shielded from the company's liabilities, with exceptions such as fraud.
  • On October 26, Rep. James Comer asked Delaware's governor for documents on safety assessments and industry involvement.

Delaware, a US state whose corporate taxes and fees bring in more than $2 billion a year, about a third of its revenue according to TechSpot, is weighing a new kind of business entity that an AI agent would run day to day. The idea has now reached Congress: on October 26, 2026, House Oversight and Government Reform Committee chairman James Comer, a Kentucky Republican, wrote to Delaware Governor Matt Meyer questioning the project.

What Delaware is considering

According to reporting summarised by TechSpot, a draft bill would create an "Artificial Intelligence Company", or AIC. It could pursue any lawful purpose except banking, and an AI agent would make business decisions without routine human supervision. The concept would first be tested in a 30-month regulatory sandbox.

Each AIC would have a single member, a person or an organisation, responsible for keeping it adequately funded. That member would generally not be liable for the company's debts or harms, except in cases such as fraud, deliberate law-breaking, or knowingly letting the AI act outside the company's stated purpose. Supporters argue that giving AI agents a legal identity would make their actions traceable and give victims a clear party to sue.

The proposal is not law. TechSpot cites local outlet Spotlight Delaware as reporting that a bill would likely be introduced next year. The work traces back to a state AI commission that, in a December 2025 report, described a subcommittee drafting a sandbox framework for agentic AI in corporate governance.

Why Congress is asking questions

Comer's committee says the liability shield could let AI firms and investors escape responsibility for reckless conduct. It also criticises the lack of transparency about which companies contributed to the proposal, noting that the December 2025 report did not mention any assessment of safety risks or consultation with independent experts. According to TechSpot, Comer requested documents including safety assessments and details of industry involvement, and warned that the inquiry could inform federal legislation limiting state AI initiatives. The governor's office has said it is reviewing the letter.

What to watch

The plan raises a basic legal question: who answers when an autonomous agent causes harm. Critics quoted in the press, including a UC Berkeley law professor cited by the Financial Times, call the proposed shield too generous. Whether Delaware presents a bill, and how it handles liability, will be closely watched by other states and by Congress.