Delaware is advancing a draft law to create “Artificial Intelligence Companies” (AICs) that would be operated entirely by AI agents, allowing owners to delegate day‑to‑day management, contract signing, litigation, and asset handling to autonomous software while shielding human members from most liability except fraud or willful lawbreaking. The proposal, co‑authored by Secretary of State Charuni Patibanda‑Sanchez and Norm Ai’s CEO John Nay, establishes a 30‑month regulatory sandbox to pilot “agentic commerce” under supervision, with capital tied to liability and the ability to shut down systems. Delaware, home to about two‑thirds of Fortune 500 firms and 334,461 new entities in 2025, expects additional revenue from formation fees for AICs. U.S. Rep. James Comer (R‑Ky.) has requested documents, warning the AIC model could be abused to evade responsibility for reckless actions and prompting possible federal limits on state AI regulation. The draft bill defines an AIC as a separate legal entity managed by an AI agent, with a single human or corporate member responsible for capitalization and activity logging, and explicitly states that the member is not liable for acts taken by the AI unless used for fraud or intentional violations. Critics, including UC Berkeley’s Andrew Baker and York University’s Martin Petrin, argue the liability shield is overly generous and could trigger a regulatory race to the bottom, urging a federal framework to address safety concerns.