Why should Delaware care?
Artificial intelligence forms the backbone of the world’s hottest industries today. Its use also raises questions about whether this technology can be used safely and for the benefit of society. Delaware officials now say they may set up a testing ground for AI agents to run entire companies.
A Delaware commission that has been studying business uses of artificial intelligence proposed a bill earlier this month that would temporarily ease state regulations for companies implementing the fast-growing technology.
The proposed bill would create a testing ground for companies to use so-called AI agents to autonomously complete business tasks normally performed by humans. AI agents will oversee entire business operations under the umbrella of a new type of organization called an artificial intelligence company (AIC).
Supporters argue that the resulting regulatory “sandbox” would allow Delaware to test how autonomous AI businesses operate and provide lawmakers with critical data to craft rules governing the use of autonomous AI within the state’s prominent corporate franchise.
The bill is likely to be introduced in next year’s General Assembly.
This month’s draft proposal comes after tech companies have poured billions of dollars into the AI industry in recent years, much of which was used to build data centers across the country to power the technology.
The bill also comes at a time when Delaware is trying to maintain its status as a prominent legal home for businesses despite attacks on the brand by the likes of the world’s richest man, Elon Musk.
John Mark Zeberkiewicz, the bill’s lead author, said the measure could allow AI agents to engage in nearly any business activity, from providing coding services to entering into contracts and even filing and defending lawsuits.
He also pointed out that the owners of new artificial intelligence companies aim to protect themselves from legal liability for actions that AI may take.
“Anything a company can do, AIC can do as a company,” said Zeberkiewicz, an attorney at Richards, Leighton & Finger in Wilmington.
But exactly how AI will work in the proving ground remains unclear, reflecting the growing uncertainty surrounding the broader AI industry.
Asked whether an AI company could take over the entire operation of a high-risk business like a small nuclear power plant, Zebarkiewicz said with a laugh: “Fair enough.”
But he said he believes that’s not what real companies do.
Sandbox proponents argue that Delaware cannot wait for such questions to be fully resolved because companies are already deploying AI agents to perform autonomous tasks. They argue that the law would allow states to set legal standards before the technology becomes more widespread.
At a June meeting of the Delaware commission, which has been considering the issue for the past year, chairman Patrick Callahan said investment bank Morgan Stanley had already given AI agents access to its stock trading platform.
He said the future the commission is planning is “actually here and now, and it’s being delivered through one of the largest financial institutions.”
AI agents are already making transactions, but there is no clear party who is responsible if something goes wrong, Callahan said.
“Technology companies will operate regardless. It’s a problem for ordinary Delawareans and the other parties to the transaction,” he said.
It was not immediately clear whether large companies or lobbying groups had asked Delaware to propose an AI testing site. Still, in April, Mark Cuban, a prominent billionaire, posted on social media site
In response to the post, Gov. Matt Meyer said, “That’s right!” Next, we provide a link to Delaware’s AI Sandbox Initiative.
Who is responsible?
The incentive for companies to join Delaware’s proposed regulatory sandbox would be to test autonomous entities with liability shields, Zeberkiewicz said.
“It’s just like any other limited liability company. It’s created to ensure that the owners of the business are not automatically liable for the entity’s debts and obligations,” he said.
Any organization incorporated in Delaware can apply to become part of the testing ground, officially called the “Artificial Intelligence Regulatory Sandbox.” In its filing, the company may ask a future oversight board for exemption from Delaware regulations, regulatory provisions and license agreements.
A future commission consisting of Delaware’s secretary of state, attorney general and others would then accept or deny the test site application.
Alan Kyle, a member of Delaware’s AI Commission, said he cannot predict which regulatory exemptions applicants will request. Callahan said the same thing.
“When people apply and say, ‘This regulation mandated by Delaware is preventing us from doing business, and we want to find an exemption from that,’ we’ll know about it,” Callahan said.
State Rep. Cindy Romer, a member of the Sandbox subcommittee, told Spotlight Delaware that the proposal will not be considered by the General Assembly until next year.
What is an AI agent?
Generally speaking, when most people think of AI, they think of generative AI, including famous chatbots such as OpenAI’s ChatGPT, Anthropic’s Claude, and Google Gemini.
Sunita Chandrasekaran, associate professor of computer and information science and director of the First State AI Institute at the University of Delaware, said that for larger, more complex tasks, generative AI is prompted multiple times and requires many “iterative interactions.”
“The first iteration doesn’t get you what you want,” Chandrasekaran said.
Chandrasekaran said that while generative AI answers questions, AI agents are “doers.”
“You’re not just asking a question, you’re asking someone to do a job for you,” Chandrasekaran said.
Among Chandrasekaran’s ongoing research is a project to build an AI agent that writes code. The agent can generate about 10,000 codes within minutes, but Chandrasekaran said this cannot be done manually.
“It would probably take several years to write this much code,” Chandrasekaran said. “The agent is now giving me more output than a human can do in a given amount of time.”
Many companies have announced job cuts related to new AI efficiencies. Some specifically cited new AI agent programs to supplement customer service, support engineers, and management positions.
But while AI agents have increased efficiency, they have also introduced new risks to businesses.
Kevin Bhimani, a software engineer on Chandrasekaran’s team, said the risks could range from sending an email to the wrong person to deleting an entire database to disclosing data to the wrong person.
“You could potentially hack into the system and access data that you didn’t intend to access,” Bhimani said.
Why Delaware?
At a meeting of the Delaware AI Commission this month, Delaware Secretary of State Charuni Patibanda Sanchez said that if testing goes well, the program would put Delaware on the “state of the art in technology.”
And just as states impose fees on forming LLCs, Delaware would benefit from fees paid for forming artificial intelligence companies, Patibanda-Sanchez said.
Currently, the collection of taxes and fees on the more than 2 million businesses that maintain a legal presence in Delaware contributes more than $2 billion to the state budget, or about one-third of all revenue.
“If a new organizational form is created out of that, that would be a really big benefit to the state’s revenue,” Patibanda Sanchez said at the June meeting.
Patibanda-Sanchez said that if AI agents prove to be a beneficial and responsible use, “we would like to be able to move the AIC law out of the sandbox and into state law.”
“That’s the whole reason to test it, instead of trying it for the first time without knowing what’s going to happen,” Patibanda-Sanchez said.
