Delaware’s push began with a broader wager: that agentic AI is coming to commerce whether lawmakers welcome it or not. In July 2025, Gov. Matt Meyer signed a resolution creating a regulatory-sandbox framework for testing novel technologies using agentic AI. The state’s AI Commission subsequently began work on a corporate-governance sandbox and potential legislation.¹

By 2026, the proposal had sharpened into a new entity: the Artificial Intelligence Company, or AIC. Under a draft bill, an AI agent could manage a company’s affairs, enter transactions, hold property, sue or be sued, and conduct nearly any lawful business other than banking. The proposed 30-month sandbox would test that model before lawmakers decide whether to make it permanent.¹

The architects’ case is that formal recognition is safer than pretending autonomous systems cannot act. Delaware Secretary of State Charuni Patibanda-Sanchez and Norm Ai chief executive John Nay argued that the answer is to “wrap AI in legal form”: give it a recognizable identity so conduct is “visible, traceable, and accountable.” They say the sandbox would run “in daylight, under supervision,” with capital tied to liability and a mechanism to shut an AI system down.¹

But the liability architecture is exactly what has triggered Washington’s alarm. The AIC’s single human or corporate member would fund the company and collect profits, while generally being shielded from its debts and from acts or omissions by the AI-directed business—apart from fraud, willful legal violations, or knowingly out-of-purpose conduct.¹

House Oversight Committee Chairman James Comer, a Kentucky Republican, asked Meyer for documents about the plan and warned that it could protect “powerful AI companies and their investors from responsibility for reckless harmful actions.” He said the inquiry could shape federal legislation limiting state AI regulatory initiatives.¹

The clash is now plain: Delaware sees a controlled test that could bring oversight—and fees—to the country’s corporate capital; Comer sees a legal innovation that could leave victims facing a company run by software while its human backers stand at a distance. Meyer’s office said it was reviewing the letter.¹