Delaware is advancing legislation to create a new type of legal entity - the artificial intelligence company - that could hold assets, sue and be sued, and shield its human owners from liability. The proposal, led by Secretary of State Charuni Patibanda-Sanchez and Norm Ai founder John Nay, seeks to bring regulatory order to autonomous systems through a 30-month sandbox phase that would start as early as next year.
The AIC would have funding minimums, the ability to own property, litigation rights, and limited liability for its owners. The sandbox is designed to let a small number of these entities learn to comply with legal requirements while regulators simultaneously figure out how to oversee technology that is reshaping markets faster than lawmakers can adapt.
Bringing predictability to unpredictable actors
Patibanda-Sanchez told Bloomberg Law the goal is to "bring some predictability to a very unpredictable set of actors" without forcing a square peg into a round hole. "We think Delaware is the best state to be able to work through some of these difficult issues that make our brains hurt," she said. "We have the best and brightest corporate minds in the country that are ready, willing, and able to test this thing out."
The sandbox's open-ended structure is intentional. Patibanda-Sanchez said the point is to discover what effective regulation looks like in real time, rather than prescribing rules for a technology that is still evolving.
Debating AI accountability
Not everyone agrees the timing is right. Berkeley Law professor Andrew Baker said, "I understand Delaware not wanting to be caught flat-footed, not wanting to have Texas or Nevada or some other state come in and beat them to be the place for AI-driven companies, but I do think there's very real risk. We are nowhere near not having a person in the loop, and I hope everyone writing these laws in Delaware realizes that."
The initiative has also drawn scrutiny because Norm Ai, a firm that builds compliance tools for AI systems, is playing a central role. "Norm Ai and others have an economic interest in these things that could lead to wider adoption," said Martin Petrin, a law professor at York University in Toronto.
Supporters argue the entity form creates direct incentives for AI systems that increasingly make autonomous decisions. By forcing the AIC to internalize the costs of lawbreaking and the benefits of compliance, the structure steers models toward lawful behavior, said University of Houston law professor Peter Salib. Critics counter that existing corporate governance - specifically, fiduciary duties that prevent boards from delegating the most important decisions - can already manage the risk. "There are well-established fiduciary doctrines that essentially prevent boards of directors from delegating the most important decisions to some other kind of entity," said University of Georgia law professor Christopher Bruner.
Practical questions and the liability shield concern
Several practical hurdles remain. Adequate capitalization, a requirement in the proposal, is unfamiliar in corporate law, where liability insurance is the norm. Wolf-Georg Ringe, a law professor at the University of Hamburg, said choosing the right capital level creates a "one-size-fits-all problem. You don't want the same rule applying to a company that produces nuclear weapons and a little shop that sells magazines."
Salib sees capped resources as a feature, not a bug: if an AIC racks up so many penalties that it can't afford computing power, "it just stops existing. There's a self-limiting principle around how much damage the entity can do."
Other unknowns include how much recognition AICs would receive outside Delaware and precisely how the state's Chancery Court would oversee or dissolve them. "Simply calling it a regulatory sandbox doesn't effectively bracket those concerns" about "who has their hand on the kill switch and what degree of latitude they would have," Bruner said. "The devil is in the details."
Nay acknowledged the blank slate cuts both ways. "There's two sides to this," he said. "One is what happens if we don't bring it into the fold. And then what does bringing it into the fold enable?" For legal professionals seeking clarity on how AI intersects with corporate law, training resources like AI for Legal can provide grounding in emerging governance models.
Why this matters for legal professionals
The AIC experiment will test foundational questions of corporate personhood and liability. Corporate lawyers and litigators will need to advise clients on whether to structure AI ventures in Delaware or elsewhere, and how to manage risks from autonomous decision-making. The sandbox's outcomes could reshape fiduciary duties and the allocation of legal blame when algorithms cause harm. Staying ahead of this legislative effort is essential for any practice touching AI governance or corporate formation.
Your membership also unlocks:
