"No human at the helm": Delaware plan would let AI agents run businesses

Delaware is developing a new corporate form, an Artificial Intelligence Company, that could let an AI agent manage a business with no human at the helm. Under a draft bill, an AIC could conduct almost any lawful business except banking: it could make transactions, sign contracts, hold property, and sue or be sued. A human or corporate “member” would provide capital and receive profits, while the AI handled daily operations. The member would generally be shielded from the AIC’s debts and from harm caused by its actions, with exceptions for undercapitalization, fraud, willful lawbreaking, or knowingly permitting activity outside the company’s stated purpose. Delaware proposes a 30-month supervised sandbox to test the model, keep activity logs, and give lawmakers evidence before deciding how to legislate.

The plan is being promoted as a way to make autonomous commerce visible and accountable by giving AI a legal identity, while also attracting incorporation fees and tax revenue to a state already home to roughly two-thirds of Fortune 500 companies. But House Oversight Committee chair James Comer has asked the governor for records, warning that the liability protections could let investors escape responsibility for reckless AI conduct and enable regulatory capture. He also raised the prospect of federal legislation limiting state initiatives. Legal scholars have questioned whether the proposed shield is too generous and warned that competition among states could trigger a race to the bottom; some argue a federal framework may be needed. Delaware’s effort grew from a state AI sandbox initiative and a partnership with Norm Ai, a company developing legal AI agents.