Delaware Bill Would Let AI Run a Company—Congress Balks

What Delaware's AI company bill actually proposes

The draft creates a new corporate form: an Artificial Intelligence Company, or AIC, where an AI agent makes business decisions instead of a human executive team.

Delaware already sells itself as the easiest place in America to incorporate. Low or zero tax exposure depending on where sales happen, specialist courts, a body of case law that lawyers know inside out. The AIC bill extends that logic to the next logical, if uncomfortable, step: a company whose decision-maker is software.

The bill sets up a 30-month regulatory sandbox. During that window, an AI agent would run the business. The company could still be sued, and it could still sue others. Banking is explicitly carved out, so no AI is getting a charter to take deposits.

One human is still required. That person is the legal anchor, not the operator. The AI makes the calls.

Why Congress is worried about an AI liability shield

Rep. James Comer, chairman of the House Oversight Committee, sent a letter to Delaware Governor Matt Meyer warning that the AIC structure could become a legal shield for the AI industry.

According to the Financial Times, Comer wrote that he is concerned the proposal "could be used to protect powerful AI companies and their investors from responsibility for reckless harmful actions." He is asking for safety assessments and has raised the prospect of federal legislation if the state's answers do not satisfy him.

The core of the objection is simple. Corporations already shield owners from personal liability, but those owners are humans who can be embarrassed, fined, or jailed. An AI agent has no such exposure. If the agent makes a destructive decision, the bill's logic points the finger at the software.

Andrew Baker, a law professor at the University of California, told the FT the bill may be an attempt to give Delaware-registered businesses "legal certainty" around agentic AI. His read on the likely outcome is less flattering: "an overly generous liability shield which seems unwarranted without some underlying policy rationale that hasn't been developed or justified well."

The Mark Cuban argument for moving first

Not everyone sees the bill as a loophole. Some in business and AI argue that if a US state does not create a legal home for AI-run companies, the business will simply incorporate somewhere else.

Mark Cuban made the case earlier this year, arguing that any governor who wants to grow state revenue should let "anyone to incorporate using an AI agent and stablecoins." His reasoning was blunt: first to market makes a killing.

That is the tension Delaware is sitting in. The state has spent decades as the default home for American corporations, and it is watching agentic AI turn from demo to deployable. Sitting still has a cost too.

The sandbox is the compromise. Thirty months to see what an AI-run company actually does before writing permanent rules. Whether a sandbox can contain a liability question this large is the part nobody has answered.

What it means for you

For now, nothing changes for founders or investors outside Delaware. The bill is a draft, the sandbox is a proposal, and no AIC has been formed. Companies incorporated in Delaware remain governed by the existing rules.

If the bill passes, the first practical question for anyone using the structure will be insurance and contracts. Counterparties in the US, UK, and EU will want to know who is personally on the hook when the agent gets it wrong. Congress is already signaling that answer may end up being written in Washington rather than Dover.

Delaware's AIC bill is a bet that the first state to give AI agents a legal home wins the next wave of incorporations. Congress is betting that the liability shield inside it is too generous to survive contact with Washington. The 30-month sandbox will be the test.