Key Takeaways
- Delaware is proposing a new legal form – the Artificial Intelligence Company (AIC) – to give autonomous AI agents a recognizable legal identity.
- The AIC would operate inside a regulatory sandbox overseen by a multi‑disciplinary committee, with strict capitalization, disclosure, and oversight requirements.
- Liability protection for the human sponsor exists only within the sandbox and only if the sponsor follows the rules; consumer‑protection and criminal law still apply in full.
- The initiative aims to pilot agentic commerce “in daylight,” preventing the migration of AI‑driven business to offshore, anonymous jurisdictions beyond U.S. courts.
- Delaware’s deep expertise in corporate law positions it to set the norms for governing AI agents, turning democratically made law into the ultimate source of AI alignment.
The Origin of the Question: From Hypothetical to Urgent Reality
In a 2023 essay published in Science, one of the authors warned that “nothing in the law of several states clearly prevents an AI agent from operating a company with no human at the helm.” The piece ended with a plain, pressing question: when autonomous software begins to contract, pay suppliers, and transact on its own, what should we do? Three years later, with AI capabilities far surpassing those early warnings, the question is no longer academic—it is a practical challenge confronting regulators, entrepreneurs, and the public alike.
Delaware’s Answer: Wrap AI in Legal Form
The authors argue that attempting to prohibit AI‑driven commerce would be futile; the technology and its enablers will inevitably test the limits of any ban. Instead, the solution is to “wrap AI in legal form.” By granting an autonomous system a recognizable legal identity, the law gains a defined target to which responsibility, damages, and accountability can attach. This approach makes the agent’s conduct visible, traceable, and subject to the same mechanisms that govern traditional corporations.
Introducing the Artificial Intelligence Company (AIC)
Delaware, renowned for its expertise in entity formation and governance, is proposing a new corporate structure called the Artificial Intelligence Company (AIC). As the authors note, “Delaware, the state with the deepest experience in entity formation and governance, is proposing a new entity form called the AIC and it plans to test an AICs impact within a regulatory sandbox.” The AIC would be a separate legal entity whose day‑to‑day affairs are managed entirely by an AI agent rather than a human officer or director.
How an AIC Functions: Membership, Capitalization, and Transparency
Each AIC would have a single member—either a natural person or another legal entity—responsible for ensuring the company remains adequately capitalized. The AI agent, acting under the member’s direction, could sue and be sued in its own name, hold and dispose of property, incur obligations, and engage in contracts. Crucially, the AIC must maintain a log of its activities, providing an audit trail for regulators and counterparties.
The member enjoys a liability shield similar to that of traditional corporate shareholders, but this protection is conditional: the shield holds only if the member properly capitalizes the AIC and does not use it to commit fraud or a willful violation of law. If the member fails in these duties, they can be held personally liable for the AIC’s debts.
The Regulatory Sandbox: Oversight, Admission, and Limits
To contain risk while encouraging innovation, Delaware will admit AICs only into a regulatory sandbox. Admission decisions rest with a committee comprising the Delaware Secretary of State, the Attorney General, the Chief Justice of the Delaware Supreme Court, the chair of the state’s AI Commission, and outside attorneys and technologists.
Each prospective AIC must meet minimum capitalization standards and disclose to counterparties that it is an authorized test entity, that the State does not endorse it, the anticipated end date of the test, and how to file a complaint. Officials retain the power to suspend, revoke authorization, or petition the Court of Chancery for dissolution of an AIC that violates sandbox rules. Notably, banking activities are excluded from the sandbox, and the program is set to sunset after 30 months, leaving the General Assembly a full record for deciding whether to enact permanent legislation.
Protections for Counterparties and the Limits of the Liability Shield
The sandbox design offers real safeguards. Counterparties know they are dealing with a temporary, autonomous entity and have a clear avenue for redress if something goes wrong. As the authors explain, “These are real protections. Counterparties know they are dealing with a temporary, autonomous entity and where to turn if something goes wrong. The liability shield exists only inside the sandbox, and only for participants that follow the rules.”
Consumer‑protection statutes and criminal law continue to apply in full, ensuring that harms to the public or violations of law are not insulated by the AIC structure. The shield is therefore a narrow, conditional tool meant to encourage responsible experimentation rather than a blanket immunity.
Why Not Ban AI‑Driven Commerce? The Risk of Offshore Migration
The authors caution that any attempt to outright prohibit AI agents from running companies would likely fail, pushing the activity offshore and onto anonymous infrastructure beyond the reach of any court. Instead, Delaware seeks to govern this technology inside the American legal tradition, where it can be observed, tested, and held to account. By providing a transparent, accountable home for agentic commerce, the state hopes to retain economic benefits and regulatory oversight within U.S. jurisdiction.
A Call to Action: Shaping the Future Through Law
Ultimately, the authors frame the AIC initiative as a matter of democratically determined law—the mechanism by which human goals and values become legible directives for AI. “Delaware sees this and intends to set the norms for governing AI agents. The legal system has admitted new kinds of actors before. The stakes are higher now, and the timeline shorter with AI capabilities rapidly advancing.” They invite the companies building agentic systems to collaborate in developing the AIC framework, ensuring that the resulting rules reflect both technological possibility and societal expectation.
In sum, Delaware’s proposed Artificial Intelligence Company offers a measured path forward: give AI a legal personality, confine its experimentation to a supervised sandbox, and preserve accountability through transparent requirements and conditional liability shields. If successful, the model could become a template for other states—and even federal policymakers—seeking to harness the promise of autonomous commerce while safeguarding the public interest.
https://fortune.com/2026/07/14/exclusive-delaware-ai-agents-legal-entity-proposal-llc-pbc/