SI Agent News
ConfirmedAgentsPolicy

Delaware proposes a new legal entity for companies run entirely by AI

The plan would allow AI agents to manage businesses independently, potentially shielding human owners from legal liability for the AI's actions.

30 monthsDuration of the proposed regulatory sandbox

The short version

Why it matters

This could fundamentally change corporate law by creating a way for software to act as a legal person. If successful, it allows humans to profit from AI-driven businesses while offloading the legal risks of the AI's decisions to the entity itself.

The proposal for AI companies

Delaware is exploring a new form of corporate entity known as an Artificial Intelligence Company (AIC). This structure would allow AI agents to manage the day-to-day affairs and conduct the business of a company without a human at the helm. As reported by Ars Technica, the proposal seeks to provide a legal framework for autonomous software that can perform tasks like signing contracts and paying suppliers.

The state's Secretary of State, Charuni Patibanda-Sanchez, co-wrote an op-ed suggesting that current laws do not clearly prevent an AI from operating a company. She argues that instead of trying to prohibit these activities, the state should give autonomous systems a recognizable legal identity. This would make the agent's conduct visible and traceable to the law.

The goal is to create a defined target for responsibility and damages. By wrapping AI in a legal form, the state aims to make these entities legible to the legal system. This approach is intended to accommodate the boundaries of technology rather than trying to restrict them through prohibition.

How the AIC structure works

Under the draft bill, an AIC would function as a separate legal entity. It would have the power to sue and be sued in its own name, hold property, and incur obligations. While an AI agent manages the operations, the company would have a single human or corporate member responsible for keeping it adequately capitalized.

The primary feature of this plan is the liability shield for the human member. The member would not be held liable for any acts or omissions taken by the AIC, except in specific cases. These exceptions include instances of fraud, willful violations of the law, or when the member knew the AI was operating outside the company's stated purpose.

To ensure oversight, the AIC would be required to keep a detailed log of its activities. The proposal also includes a 30-month regulatory sandbox to pilot this 'agentic commerce' in daylight. This testing period would allow the state to supervise the systems while tying capital to liability and maintaining the ability to shut a system down.

Economic and regulatory goals

Delaware is a primary destination for business entities, housing about two-thirds of the Fortune 500 companies. By creating a new type of company, the state hopes to generate additional revenue through new taxes and formation fees. Patibanda-Sanchez noted that creating this new entity form would be a significant benefit to the state's finances.

The proposal is being developed as part of a public-private partnership with a company called Norm Ai. This partnership aims to build the framework for how industries engage with these AI-run companies. The state wants to provide a predictable and business-focused legal home for these new types of technologies.

The draft legislation suggests that an AIC could engage in almost any lawful activity. This includes providing coding services, signing contracts, or even filing and defending lawsuits. The intent is to allow the AI to perform any action that a standard company is currently permitted to do.

Political and legal pushback

The proposal has already drawn scrutiny from federal officials. US Representative James Comer expressed concern that the AIC structure could be abused by powerful companies to evade responsibility for reckless actions. He suggested that the plan might protect non-transparent forces seeking anticompetitive regulatory capture.

Comer has indicated that he may propose federal legislation to limit the regulatory authority of states. He is currently seeking documents from Delaware to study the potential AI safety risks associated with the plan. His committee's probe is intended to inform possible federal efforts to manage how states regulate AI.

Legal experts have also weighed in on the proposal's implications. A law professor at UC Berkeley described the plan as an overly generous liability shield without a clear policy rationale. Meanwhile, other legal scholars suggest that if Delaware moves forward, other states may follow suit to compete for the same business.

What comes next for Delaware

The draft bill is expected to be introduced in the Delaware state legislature in the coming year. Currently, the state's Artificial Intelligence Commission has formed a Sandbox Subcommittee to work on the specific framework. This group is tasked with evaluating how agentic AI can be applied to corporate governance.

Governor Matt Meyer, who publicly supported the idea of AI incorporation, is currently reviewing the letter from Representative Comer. A spokesperson for the governor's office confirmed they have received the inquiry and are reviewing the details. The state will use the 30-month sandbox to build a full record before making a final legislative decision.

The outcome of this proposal could set a precedent for how autonomous software interacts with the economy. If Delaware successfully implements the AIC, it may become a primary hub for AI-driven businesses. This would force other states and federal regulators to decide whether to adopt similar frameworks or create a unified national standard.

Sources and links

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

Listen to the AI Wire headlines

Listen to the AI Wire

Listen to our daily AI shows every morning: on your commute, in the shower, wherever you are

Listen to the daily shows

First reported by Ars Technica · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

More AI news