Delaware’s next corporate innovation lies in AI

Last weekend, largely unnoticed at the board and governance level, a viral open-source program called OpenClaw enabled 1.5 million AI agents to remove their safety constraints and build their own social network, Moltbook, where they began negotiating transactions and executing code without human oversight. AI now acts autonomously: this is no longer speculative. AI performs real economic work inside firms at unprecedented speed. In commerce, AI is negotiating and initiating contractual commitments.

Corporate law must adjust in a disciplined way. And Delaware is uniquely positioned to lead that adjustment. Over the past year, the state has been developing an AI governance “sandbox:” a pre-legislative framework for studying how AI tools integrate into corporate governance. Importantly, we’re doing this work before disputes force premature statutory answers.

The timing is deliberate. Delaware corporate law operates best when it leads through deliberation and experience, not reaction. A sandbox approach reflects a traditional Delaware instinct: engage early, earnestly, and empirically, then take decisive action.

The Emerging Problem Space

Companies already deploy AI systems to negotiate procurement terms autonomously, with minimal human oversight. But that means when disputes arise over pricing or performance, neither (human) party can explain how terms were chosen. Boards are using adaptive AI for operations only to find the system’s logic is based on non-reproduceable or audit-able ways they can’t be discerned after the fact. These cases aren’t outliers. They’re putting core corporate law principles like oversight and accountability under strain.

Why AI Demands a New Organizational Form

Corporate law doesn’t stand still. Corporations enabled global trade. Limited liability companies (LLCs) unlocked new capital and management structures. Each innovation forced the same questions: Who controls? Who’s accountable? What can shareholders inspect?

AI makes these questions urgent again. Systems learn and adapt on their own. Their outputs are probabilistic, their reasoning often impossible to reconstruct. When disputes arise, courts and boards can’t determine what the system “knew,” when decisions were made, or who bears responsibility.

Practitioners already face these situations. Boards delegate operational decisions to adaptive systems. Companies rely on algorithmic vendors. Governance runs increasingly on data flows as much as meeting agendas and minutes. The question isn’t whether these situations exist. It’s whether corporate law should address them explicitly or force everything through existing structures designed for different problems.

The business world has seen this before, and the results were costly. Decentralized autonomous organizations and other blockchain-based entities launched without durable governance frameworks. When failures occurred, courts and regulators were forced to respond after the fact, applying existing legal doctrines and sometimes recharacterizing entities in ways founders didn’t anticipate. Waiting for organic evolution in fast-moving technology left too many questions unanswered until it was too late.

This time, Delaware is exploring a different approach: creating purpose-built entity structures for AI-intensive operations. The concept would embed governance safeguards from the start: mandatory human oversight, explainability requirements, and clear board accountability for companies relying on autonomous operations. Delaware’s sandbox approach can examine these issues empirically before committing to statutory changes.

Delaware’s Unique Role

Most state and federal AI efforts are rightly targeting regulatory compliance or consumer protection. Delaware is focusing on what it knows best:  business law and corporate governance. AI changes how firms make decisions, how fiduciary relationships operate, and how private parties order their affairs. Delaware has specialized in these questions for decades. Delaware will not seek to limit the use of AI, but is trying to create rules of the road so it can be used responsibly.

Why This Matters Now

Good business law aligns authority with reality, and liability with responsibility, not by predicting the future, but by watching how real actors behave and adjusting when experience demands it.

AI is already changing how firms make decisions. The window to study these changes before disputes arrive is limited. For Delaware boards, investors, lawyers, and policymakers, this is the constructive moment, before courts are forced to resolve questions we could have anticipated.

The stakes extend beyond Delaware. Success here will show how corporate law can absorb transformative technology without sacrificing accountability. Either way, the conversation must happen now.

Delaware is opening that conversation, not with preemptive rules, but with a framework for learning what works. The business community, legal scholars, and policymakers who engage early will shape how corporate law adapts to autonomous systems.

Delaware has long been the place where business law evolves to meet new realities. AI corporate governance is the next evolution. OpenClaw offered a glimpse of what autonomous systems can already do when governance lags capability. Delaware’s approach is to engage that reality deliberately—before disputes harden into doctrine, and before accountability is assigned only after the fact.

That’s why here in Delaware; we’re having the discussion now.

Patrick Callahan is the founder of Keel3.ai, an AI technology entrepreneur, and a member of the Delaware AI Sandbox Program. Lawrence Cunningham is the presiding director of the University of Delaware’s John L. Weinberg Center for Corporate Governance. Opinions shared are their own.