Microsoft AI Chief Suleyman Calls Delaware's AI-Agent Company Plan ‘Deeply Concerning’

Microsoft AI chief executive Mustafa Suleyman has publicly pushed back on a Delaware legislative proposal that would let autonomous AI agents legally manage and operate corporate entities, calling it “deeply concerning.” He made the comment on X, turning a state-level drafting exercise into a national argument about AI governance.

The disputed clause is explicit: a Delaware Artificial Intelligence Company would be a separate legal entity whose day-to-day affairs are managed by an AI agent rather than a person, and at the agent's direction the company could sue and be sued in its own name, hold and dispose of property, and incur obligations. Suleyman quoted that language directly and wrote that “we should be very cautious here.”

Under the draft bill that may go before the state legislature, an AIC could pursue any lawful commercial purpose, for-profit or not, with banking as the only explicit exclusion. An AI agent could be formally designated to run day-to-day affairs, enter contracts, make commercial transactions and direct strategy with no human at the helm. The framework also envisions a 30-month regulatory sandbox to pilot the structure.

Delaware has its own reasons to try. It is the preferred home for US incorporations, and the draft is meant to bring additional state revenue through corporate filing fees and taxes while keeping a new kind of AI-native company domiciled in the state. To supporters that is institutional innovation; to critics it outsources accountability to a line of code.

Washington responded quickly. Representative James Comer, the Kentucky Republican who chairs the House Committee on Oversight and Government Reform, sent a formal inquiry to Governor Matt Meyer last month demanding documents on the AIC framework and its safety risks. He warned the proposal could shield powerful AI companies and their investors from responsibility for reckless harmful actions while protecting non-transparent actors seeking anticompetitive regulatory capture, and said federal legislation could be introduced to preempt state AI efforts directly.

The real question is not capability but legal personhood. Once an agent can sign contracts, hold property and litigate in its own name, who bears liability stops being something a product manual can answer and becomes a question for corporate and tort law — which is why a tech executive's social media post drew a formal inquiry from a congressional committee chairman.

Three things to watch: whether the draft actually reaches the state legislature, how tightly the 30-month sandbox is written, and whether Washington moves to preempt. For companies building agent products, Delaware's choice is likely to become the template other states copy.