An AI would run the company while humans supplied the capital. Who would answer for the harm it caused? On October 26, US Representative James Comer asked Delaware Governor Matt Meyer for records about the state’s proposed Artificial Intelligence Company structure.
The Republican chair of the House Oversight and Government Reform Committee warns that the plan could shield AI companies and their investors from responsibility for reckless, harmful actions. His letter requests drafts, outside agreements and communications about safety risks and liability limits. It sets a November 9 deadline.
Comer also criticizes a lack of transparency in how the proposal was developed. That is the committee chair’s assessment, not an established legal violation. He says the inquiry could inform possible federal legislation limiting state AI regulatory initiatives.
The published draft shows that the dispute is not about unlimited immunity. It envisages a separate company managed by an AI agent. Its member would not be personally responsible for company liabilities solely because of that ownership role. Liability protection also depends on adequate capitalization. Exceptions include fraud, willful legal violations and allowing activities beyond the company’s stated purpose when the member knew or should have known about them.
The draft also includes controls. The AI manager would have to document its actions in signed records. Under specified conditions, the sandbox administrator could suspend or revoke participation and order a company wound up. The regulatory sandbox would last 30 months. These are draft provisions, not evidence of an enacted AIC law.
Meyer’s office told Ars Technica that it had received the letter and was reviewing it. Whether Delaware will change the proposal or what records it will provide remains open. The dispute concerns future oversight and accountability for AI-run businesses, not a newly established case of harm.