The state of Delaware, long considered the corporate capital of the United States, finds itself at the center of a burgeoning national debate over the intersection of artificial intelligence and legal liability. A legislative proposal currently under development in the state aims to create a new class of corporate entity known as an Artificial Intelligence Company (AIC). Under this framework, autonomous software agents would be granted the legal authority to manage business affairs, enter into contracts, and conduct transactions without the oversight of a human director. However, the initiative has drawn sharp criticism from federal lawmakers, who warn that the proposal could inadvertently create a regulatory vacuum where corporate accountability evaporates.
U.S. Representative James Comer (R-Ky.), Chairman of the House Committee on Oversight and Government Reform, formally intervened in the matter yesterday. In a letter addressed to Delaware Governor Matt Meyer, Comer requested extensive documentation regarding the state’s efforts to establish these AI-run entities. Comer’s correspondence signaled a potential escalation, as he suggested the committee is considering federal legislation that would limit the authority of individual states to experiment with such high-stakes corporate structures. The central concern, according to the Oversight Committee, is that the AIC proposal serves as a shield for investors and powerful technology firms, potentially insulating them from the fallout of "reckless" or harmful actions executed by autonomous systems.
A Chronology of the Delaware AI Sandbox
The movement toward formalizing AI-led corporations began in earnest in mid-2025. In July of that year, Governor Matt Meyer signed a legislative resolution designed to create a "regulatory sandbox" for the testing of innovative, agentic AI technologies. The goal, as stated by the governor’s office, was to solidify Delaware’s position as a national leader in responsible technological innovation.
By December 2025, the Delaware Artificial Intelligence Commission released its annual report, confirming that a dedicated Sandbox Subcommittee was actively drafting a framework to evaluate how agentic AI could be applied to corporate governance. This period of internal development moved into the public sphere in June 2026, when reports surfaced detailing draft legislation that would allow AI agents to engage in nearly any lawful business activity—ranging from coding and contract negotiation to the filing and defense of lawsuits.
The public face of the project, Delaware Secretary of State Charuni Patibanda-Sanchez, solidified the state’s intent in a July 2026 op-ed. Patibanda-Sanchez argued that because current state laws do not explicitly prohibit AI agents from operating companies, the state must proactively "wrap AI in legal form." By granting autonomous systems a recognizable legal identity, the proponents argue, the state makes the AI "legible to law," thereby creating a defined target for legal accountability.
The Mechanism of the Artificial Intelligence Company
The proposed structure for an AIC is designed to mimic the benefits of a limited liability company (LLC) while introducing a radical departure in management. Under the current draft bill, an AIC would function as a separate legal entity. A "member"—the human or corporate backer—would provide the capital necessary for the AIC to operate. Once established, the day-to-day operations are handed over to an AI agent.
The draft legislation explicitly states that the member is generally shielded from the AIC’s debts and liabilities, provided the company remains adequately capitalized. The bill proposes a significant legal firewall: the human member is not liable for any act or omission taken by the AIC or by any person acting at the direction of the AI.
The exceptions to this protection are narrow. A member would only face liability if they use the AIC to engage in direct fraud, commit a willful violation of the law, or allow the entity to conduct business that the member knew, or should have known, fell outside the company’s stated purpose. Proponents argue this structure provides necessary clarity for developers, while critics like Representative Comer argue it creates an "overly generous" liability shield that could be exploited by bad actors to evade the consequences of algorithmic harm.
Economic Motivations and Corporate Dominance
Delaware’s push to legalize AI-run companies is inextricably linked to its status as the premier jurisdiction for American business. Approximately two-thirds of all Fortune 500 companies are incorporated in Delaware, drawn by a sophisticated and highly predictable legal environment, particularly the Court of Chancery. In 2025 alone, 334,461 new entities were formed in the state.
The revenue incentive is clear. State officials have openly acknowledged that by creating a new category of entity, Delaware can capture additional fees and tax revenue. In his social media interactions, Governor Meyer signaled his support for the concept after public encouragement from tech investor Mark Cuban, who suggested that the first state to allow AI-agent incorporation would likely see a surge in revenue.
However, this competitive drive toward "first-mover advantage" has invited skepticism from legal scholars. Andrew Baker, a professor at UC Berkeley School of Law, has characterized the proposal as lacking a robust policy rationale, suggesting that the liability protections are disproportionate to the stated goals. Similarly, Martin Petrin of York University has warned of a "regulatory race to the bottom," where states might weaken oversight standards to attract AI-focused firms, potentially necessitating a federal framework to standardize safety protocols.
The Role of Public-Private Partnerships
The development of the AIC framework has not occurred in a vacuum. The Delaware Secretary of State’s office has collaborated with Norm Ai, a private firm specializing in the development of "legal AI agents." The public-private partnership is intended to ensure that the regulatory sandbox has a functional, industry-tested foundation.
According to Secretary Patibanda-Sanchez, the 30-month sandbox period is essential for gathering data. By allowing these companies to operate "in daylight" under state supervision, Delaware hopes to create a full evidentiary record that will inform future legislation. This approach, she argues, is far superior to a total prohibition, which she believes would only drive the technology into unregulated, clandestine environments.
Implications for Federal Oversight
The intervention by the House Oversight Committee marks a significant shift in the federal government’s posture toward state-level tech regulation. Representative Comer’s letter requests detailed internal communications, drafts of the legislation, and, crucially, risk assessments regarding AI safety. The threat of federal preemption—where the U.S. government would step in to set uniform national standards—could fundamentally alter Delaware’s legislative trajectory.
The core of the conflict is a philosophical disagreement regarding the nature of corporate agency. Proponents of the Delaware model view AI agents as the next logical step in the evolution of the corporation, effectively treating software as a "legal person" capable of assuming the rights and duties of a company. Detractors, however, argue that this creates an accountability vacuum where the "human in the loop" is removed, leaving victims of algorithmic error—such as consumers, business partners, or the public—with no clear entity to sue for damages.
As the state of Delaware continues to review the House Committee’s inquiry, the broader business community remains in a state of observation. If Delaware moves forward with the AIC legislation, it will likely trigger a wave of legal challenges and could set a precedent that other states are forced to follow—or reject—in order to maintain their own competitiveness. The outcome of this dispute will not only define the future of corporate law in Delaware but will likely establish the legal standards for AI autonomy for years to come.
