A Delaware committee has drafted legislation for the AIC: an entity whose daily affairs are run by an AI agent, not a person, drafted after its chair confirmed Morgan Stanley already lets AI agents trade with no clear party accountable when one causes harm. Every board already granting agents contracting or trading authority faces that jurisdiction first. The posture required is offensive: set the delegation standard now, before Delaware does it for you.
PRIORITY 9 | SILO: JUDICIAL (DELAWARE CORPORATE LAW)
A Delaware committee has drafted legislation creating the AIC, a new entity whose daily affairs an AI agent runs instead of a person, disclosed alongside confirmation that Morgan Stanley already lets AI agents trade with no clear party accountable.
The Signal
A Delaware committee that spent the past year studying business uses of artificial intelligence has drafted legislation to create a new corporate structure: the Artificial Intelligence Company, or AIC, an entity whose day-to-day operations are run by an AI agent instead of a person. Under the proposal, first reported by Spotlight Delaware on June 30, 2026 and detailed by the bill’s principal drafter, John Mark Zeberkiewicz of Richards Layton and Finger, an AIC could sign contracts, provide services, and file or defend lawsuits under AI direction, inside a state-supervised structure called the Artificial Intelligence Regulatory Sandbox.
The bill is not yet law. State Representative Cyndie Romer, who sits on the sandbox subcommittee, told Spotlight Delaware the General Assembly is not expected to take it up until next year. Committee chair Patrick Callahan told fellow members in June that Morgan Stanley already grants AI agents access to its stock trading platforms, with no clear party accountable when one causes harm.
The Evidence
The oversight structure is specific. A future admissions committee, composed of the Delaware Secretary of State, the Attorney General, and other officials, would review each AIC application and could grant exemptions from named Delaware regulations. Zeberkiewicz described the core mechanic in familiar terms: the AIC works “like any limited liability company,” shielding the owner from the entity’s debts, provided the owner keeps it capitalized and does not use it to commit fraud.
Secretary Patibanda-Sanchez told the AI Commission the same month that a successful pilot could become a durable source of entity-formation revenue on top of the 2 billion dollars Delaware already collects annually from its corporate franchise, and that the state intends to move the framework from sandbox into permanent law if the test succeeds. In a July 14, 2026 Fortune commentary co-authored with Norm Ai founder John Nay, she confirmed the design: single-member entities, mandatory capitalization and activity logs, and Court of Chancery authority to dissolve any AIC that fails to comply.
The Strategic Implication
Defensive Risk. General counsel and audit committee chairs at any company already granting AI agents authority to trade, contract, or transact, the pattern the committee chair named directly at Morgan Stanley, are exposed first. What specifically breaks: an agent transacting today, ahead of any AIC-style capitalization and logging structure, sits in the same accountability gap Delaware’s own committee chair said has no clear responsible party. Before the General Assembly takes up this bill next year, and before the next 10-K certification cycle, the audit committee needs a documented answer to who is accountable for every AI agent already contracting, trading, or transacting on the company’s behalf. The defense move: commission a one-page inventory of every AI agent with contracting, payment, or trading authority, and the named human or entity accountable for each.
Offensive Advantage. Boards that engage now, while the bill is still committee-drafted and Delaware is explicitly building industry input into the AIC framework, gain a seat in shaping the standard rather than reacting to it later. The specific advantage: a company that can already show a capitalized, logged, board-approved delegation architecture for its AI agents is the natural first applicant once the Artificial Intelligence Regulatory Sandbox opens, gaining a state-recognized structure for autonomous operations before competitors have one. Ahead of the bill’s introduction next year, the board should direct management to identify which business units would benefit from operating inside the sandbox. The move: assign one executive to track the bill through the General Assembly and report back at the next scheduled board meeting.
Delaware is not waiting for boards to define this. A state committee already has, informed by the fact that Morgan Stanley’s AI agents are trading today with nobody agreed on who answers for them. The Governance Boundary Principle applies here in its sharpest form: a board that waits for a state committee, a General Assembly, or a court to define where its AI agents’ authority ends is not exercising oversight, it is inheriting somebody else’s definition of it. Boards that write their own delegation standard before this bill reaches a vote will be negotiating with Delaware. The ones that wait will be complying with it.
Board chairs and audit chairs: Take the Board Fiduciary AI Stress Test at touchstonepublishers.com/board-fiduciary-assessment