In Delaware, a state that has long been a mecca for business, a new type of legal entity has been proposed: the Artificial Intelligence Company (AIC). In these companies, key management decisions will be made not by a human, but by an AI agent. This is not science fiction, but a very specific legislative initiative.
AICs will receive full legal personality: they will be able to enter into contracts, own property, file lawsuits, and be defendants. But there is a nuance: such companies will only be able to operate within a regulatory sandbox—an experimental legal sandbox. It will be in effect for 30 months, and outside of it, AIC activities will be impossible.
Why is this important? AI agents are already becoming part of the agentic economy. They write letters, process data, make decisions—all based on algorithms. However, the lack of a legal form makes their actions legally ambiguous. The bill proposes to put them into clear frameworks so that accountability becomes traceable. For example, if an AI agent enters into an unfavorable contract, responsibility for this will be placed on the AIC, not on the system owner.
Of course, this is not unconditional freedom. Each AIC must have at least one participant—an individual or legal entity—that is responsible for capitalization. If the company is undercapitalized or involved in fraud, limited liability does not apply. This is an attempt to avoid abuse.
The idea is already generating interest. On the horizon are payment protocols and networks for agents, databases of court decisions for their work, and even a potential market for legal services for AI. But for now, this is a cautious step that shows what the future might look like. It is possible that other countries will begin to recognize the legal personality of AI agents, and lawyers will have to deal with contracts made not by people, but by machines.
Comments
0No comments yet.