Senators on both sides of the aisle compete to propose proposals for federal regulation of artificial intelligence, but most of the attention lies in Senate Commerce Chair Ted Cruz.
The Texas Republican proposed a legislative framework last month, calling for a “light-touch” approach, but he only released one bill.
It’s not the only way he wants to advance AI development. Cruz has been lacking in getting a suspension on state AI regulations included in the recent Republican Budget Settlement Act, but he has recently said he hasn’t given up.
Meanwhile, other AI-related proposals have emerged: A “roadmap” from Senator Mark Kelly of D-Ariz would tax AI companies to deal with social and environmental harms.
Still, Cruz could be the best chance to successfully craft his agenda, given his position on the Senate Commerce Committee and his potential support for his plans from the huge tech companies that are pushing AI across products.
Play with the “Sandbox”
That framework that Cruz said was based on “five pillars” remained primarily a policy proposal, but included his bill to establish a federal regulation of AI, “sandbox.”
The bill allows businesses to allow waive federal regulations related to AI systems for up to 10 years.
The proposal has already attracted criticism from groups involved in the potential misuse of AI, giving businesses the luxury of violating civil liberties and consumer protections.
Cruz tried to address these issues when he published the framework saying that sandboxing was not a “free pass.”
“People who create or use AI must follow the same laws as everyone else,” he said. “Our laws are adapted to this new technology, and our judges regularly apply existing consumer protection, contracts, negligence, copyright laws and more to cases involving AI.”
The bill requires businesses to explain how abandoning regulations would benefit customers, efficiency, or innovation. Developers should also list foreseeable risks and explain how they outweigh the benefits. The bill requires agencies to approve applications to identify how consumers are protected.
Additionally, the White House asks Congress to amend or eliminate regulations each year, but the program shows that AI companies can “run safely.”
The sandbox bill is not universally popular with GOP. Holy said he doesn’t like the idea and the level of regulations is already low.
“The business will be doing well, as is the case. It’s a myth that businesses need this, or something else from the government,” he said. “Now, companies do whatever they want. They are the richest companies in the world.”
A wider approach
Even if it focuses on AI, the regulatory sandbox is nothing new. Texas and Utah have passed laws to establish AI sandboxes, and Delaware is planning its own program. However, Cruz’s proposals could last longer and include more companies.
This is according to Sophie Tomlinson, program director for the DataSphere Initiative, a Swiss-based research and policy group focusing on data governance. Tomlinson described the sandbox as a “safe space” and a “safe space” to test technology and practice, supporting innovation and “regulatory agility and flexibility.”
In Tomlinson’s experience, most sandboxes ran for six months to a year, and after tweaks to the program after the first round, they could be expanded by the presence of a new cohort.
Sandboxes can have different goals, Tomlinson said. Some people focus on compliance, while others create “a margin of regulation type in a controlled environment over a specific period of time.”
She highlighted the importance of clear governance regarding accountability and potential risk, including rules regarding data and intellectual property sharing within the program. She also said it would be best to be specific about what standards the sandbox deals with.
In the case of the US sandbox, she noted that the program does not close out the specific number of companies that are participating.
“The broad approach is very interesting, but at the same time it needs to make sense. It also requires monitoring and risk reduction. It also requires that regulators have practically useful learning and adapt to innovation.”
Speaking on the condition of anonymity, known as Cruz’s bill, the industry lobbyist said it was “a great way to address some of the regulatory challenges faced by AI deployers, particularly those that AI developers are more interested in regulation at the state level.
The lobbyists were also concerned that the White House science office would not have the necessary authority to implement the program. The lobbyists said the law should give power to individual institutions for sector-specific sandboxes.
Amy Bos, director of state and federal affairs at NetChoice, an industrial group that counts Amazon.com Inc. and Google LLC as members and works online at “free enterprises and free representation,” sees the bill as a good start to allow “innovators to innovate.”
“It’s actually a fundamental issue and that’s… our current regulations were written for the pre-AI world,” Boss said. “So, now I think you’re looking at AI applications. They could be blocked by decades-old rules that don’t mean so far for modern technology. So the sandbox bill gives businesses.
The sandbox proposal is less popular with people who are concerned about the risks pose by AI. The bill requires businesses to notify the sandbox program of incidents that cause harm, such as health and safety and the economy. The bill defines health and safety risks as including physical harm, death, or substantial health effects. Economic risk is defined as “specific and physical harm to a consumer’s property or assets.”
Cody Wenke, senior policy advisor to the American Civil Liberties Union, said the bill would provide “ROVING PRIMIRNIONS” for AI developers and deployers that violate key rules on consumer protection and civil rights.
“I don’t think sandboxes should be seen as some sort of medium compromise,” says Venzke. “They essentially give entities permission to be exempt from existing laws. And these are laws that apply to everyone else. And there is no good reason for us to say to artificial intelligence, “You’re special. You’ll leave prison for free.” ”
Moratrium redux?
Cruz’s bill doesn’t include a suspension on state law’s federal preemption or new regulations, but it’s possible that such a moratorium is still coming.
He said at an event last month that he “I think we’re still there and working closely with the White House,” but that there are no plans to be made public about how the moratorium will move to the floor. The revised proposal by Cruz was removed this summer from the settlement bill in a 99-1 Senate vote.
Industry lobbyists said their supporters were considering “all available options.”
“I’ve had a conversation by the end of the year looking at all the moving laws,” the lobbyist said.
BOS said they want to avoid “nightmare scenarios” in which companies navigate various state regulations regarding AI.
“I hope that the law can still move this year. I’ve heard it’s a hill’s priority,” she said. “I think it’s supported by the White House and by the administration.”
Holy opposes moratoriums and calls them “crazy” to take the lead in states that have legislated to protect their children from online exploitation and deep fur cake pornography.
“I want to protect the kids from all those things, and I want the state to be able to do that. Do you know? I mean, I think it’s a terrible idea to say, ‘Oh, you can’t protect your kids in your own condition,'” Holy said.
Venzke opposes the preemption because he doesn’t want to “evacuate” the role of states that are not federally regulated.
“The nation is a well-known, democratic institute you know, and they are already working on addressing these issues, which will help the federal government formulate what that response is.”
“Light Touch” Framework
Outside of the sandbox bill, the cruise framework promises multiple goals. Permitting AI infrastructure, open data of federal data for model training, “protecting free speech”, “unfair use of AI”, “protecting human values and dignity.”
The boss said, “The devil is always in detail, along with other bills, but we like where this is heading.”
Industry lobbyists said it was “a little quick” to know how the framework is translated into actual invoices.
“I’m really curious to see how I’m going to write laws that protect freedom of speech and defend human values in the age of AI.”
The lobbyist was also concerned about the free-speaking part of Cruz’s plan. In particular, following the path to dealing with “awakening AI” can lead to losing the possibility of a bipartisan path.
“These kinds of ideas are more about messaging than actually working with Congressional allies to advance some of these principles,” the lobbyist said.
Holy said he took a “somewhat different view” from AI’s Cruz, and was particularly concerned about workers and data use.
“I think we need to have guardrails in place that protect people and guarantee their rights,” he said.
In election year, lobbyists said even bipartisan regions of AI policy could be politicized.
“I’m pretty pessimistic in the sense that the bill has been passed and things are actually moving through Congress,” the lobbyist said.
