IA · 28 October 2026 · 4 min read
No Human at the Helm: Delaware Proposes Corporate Entities Run Entirely by AI Agents
In brief: Delaware is drafting groundbreaking legislation to establish Artificial Intelligence Companies (AICs), corporate entities that can be managed and operated autonomously by AI software without a human executive at the helm. Backers propose a 30-month regulatory sandbox to test autonomous corporate decision-making, contracts, and treasury management. However, the plan has triggered intense pushback in Washington, where congressional leaders warn that removing human leadership could allow investors to dodge liability for reckless algorithmic behavior.
by Team Mocchi's
Delaware, long recognized as the preeminent corporate haven for Fortune 500 enterprises, is preparing what could become the most consequential shift in modern business law. State administrators are drafting legislation to introduce a novel class of business organization known as the Artificial Intelligence Company (AIC)—a corporate body whose day-to-day operations and governance can be delegated entirely to autonomous AI software, without requiring a human director at the helm.
The proposal aims to bring legal clarity to autonomous software agents already executing financial transactions and negotiating contracts on the web. Yet, it has instantly sparked political friction with Washington. As reported by Ars Technica, the House Committee on Oversight and Government Reform has launched an inquiry into the state's plan, warning that the move could allow founders and investors to shield themselves from legal accountability when autonomous systems act recklessly.
A thirty-month sandbox for autonomous firms
Under the preliminary draft legislation, an AIC would be permitted to conduct any lawful business, whether for-profit or non-profit, with a single strict carveout: commercial banking. Within that perimeter, autonomous software agents could be formally appointed to oversee corporate business and affairs, executing binding contracts, issuing payments to vendors, and managing resources without obligatory human review.
To evaluate the concept while containing broader systemic risk, Delaware plans to launch a 30-month regulatory sandbox. During this trial period, algorithmic companies would operate within specialized reporting parameters. The move is designed to generate tax revenue and corporate filing fees for the state, helping Delaware sustain its status as the default jurisdiction for cutting-edge corporate structures.
Giving software legal standing
The rationale behind the draft bill was outlined earlier this year by Delaware Secretary of State Charuni Patibanda-Sanchez, who argued that current statutes in several states do not explicitly bar AI systems from operating a firm without human intervention. Confronted with agentic workflows that already cross corporate boundaries, proponents contend that outright bans are doomed to fail against decentralization.
Instead, Delaware officials argue for wrapping autonomous software into an identifiable legal framework. In their assessment, giving an autonomous agent a recognized corporate identity makes its activity legible to courts and regulators. Rather than facilitating regulatory evasion, the framework is meant to provide a clear legal entity against which damages, liabilities, and contractual enforcement can be anchored, backed by segregated corporate capital.
Washington warns of liability loopholes
Capitol Hill lawmakers, however, remain deeply skeptical. House Oversight Committee Chairman James Comer sent a formal request for information to Delaware Governor Matt Meyer, seeking all records regarding the genesis of the AIC framework and related risk assessments.
Federal investigators fear the structure will serve as an impenetrable legal shield for tech companies and venture capital firms. By offloading management decisions to autonomous code, human owners might claim immunity from liability when an algorithm engages in anti-competitive conduct, breaches consumer protection laws, or suffers severe hallucinations. Comer indicated that the probe could lead to federal legislation limiting individual states' authority to regulate or charter autonomous corporate actors.
Mocchi's take
Delaware's proposal brings into sharp focus a transition that many technology leaders still view as distant: the migration of AI agents from passive operational tools to legally consequential decision-makers. For European and Italian businesses operating under the strict guardrails of the EU AI Act and traditional civil liability frameworks, companies completely devoid of human leadership remain legally unviable for now. Nevertheless, the rapid emergence of agentic infrastructure requires us to rethink corporate IT governance immediately, establishing strict limits on automated procurement and signing authorities before autonomous efficiency turns into uncontrollable liability.