Sen. Ted Cruz’s (R-Texas) latest AI proposal puts a familiar fight back at the center of Washington’s technology debate: whether artificial intelligence should move faster with fewer regulatory barriers, or whether public safeguards should remain firmly in place as companies test new products.
The proposal, called the SANDBOX Act, would create a path for AI companies to ask federal agencies not to enforce certain laws against them while they test new AI products. Supporters describe that as flexibility for emerging technology. Critics describe it as a dangerous opening for powerful companies to avoid rules meant to protect the public.
What the SANDBOX Act would do
The SANDBOX Act is shorthand for “Strengthening Artificial intelligence Normalization and Diffusion By Oversight and eXperimentation.” It is part of Cruz’s broader push for a “light-touch” regulatory approach intended to “advance American leadership” in AI and keep “American values” at the center of the technology rather than Chinese values.
Under the bill, AI companies could apply for temporary non-enforcement of federal laws that might restrict the testing of new AI products. Those applications would need to describe known risks or harms, possible mitigation steps, and benefits that could outweigh harms.
The federal agency responsible for enforcing the relevant law would review the request and decide how enforcement should be modified. But the White House Office of Science and Technology Policy (OSTP) would also have authority to overrule decisions from independent agencies, including agencies focused on consumer protection.
The non-enforcement periods could last two years and be renewed up to four times, reaching a maximum of 10 years. Cruz’s one-pager also says Congress would be prompted to make permanent any “successful” moratoriums found to benefit the US.
Why critics see a risk to public safeguards
Critics argue that the framework could let major AI companies negotiate favorable treatment with the Trump administration while avoiding laws designed to limit harmful experimentation. The concern is not only that rules could be paused, but that the power to decide could shift toward the White House.
The Tech Oversight Project, a nonprofit tech industry watchdog group, warned that the bill could encourage “sweetheart” deals. The group suggested that Big Tech companies “donating to Trump” could gain an advantage over smaller AI firms that lack similar political leverage.
Cruz’s SANDBOX Act “would give unprecedented authority for the Trump Administration to trade away protections for children and seniors and dole out favors to Big Tech companies like Google, Apple, Meta, Amazon, and OpenAI,” the Tech Oversight Project alleged.
The bill’s text does include health and safety risks that could cause a request to be denied. Those include risks of “bodily harm to a human life,” “loss of human life,” and “a substantial adverse effect on the health of a human.” Critics still argue that the review process may not give agencies enough time to weigh those risks carefully.
The review timeline is a central fault line
Cruz’s bill requires agencies to review requests from AI companies within 14 days. Once that process begins, agencies can bring in advisory boards or working groups to assess risk, but they must reach a decision within 60 days. If they do not, the request is presumed approved.
Only one 30-day extension may be granted. That structure is one reason opponents say the bill could push agencies toward approval by default, especially when the questions involve complex AI systems and potential effects on consumers.
The proposal also includes reporting obligations after a company receives relief. AI companies would need to report within 72 hours of “any incident that results in harm to the health and safety of a consumer, economic damage, or an unfair or deceptive trade practice.”
After that, firms would have 30 days to fix the problem or risk enforcement of the law they sought to avoid. The public would also be alerted and given an opportunity to comment.
Supporters argue old rules can block useful AI
Supporters of the SANDBOX Act include the US Chamber of Commerce and NetChoice, a trade association representing Big Tech companies. Other supporters named in the source include the Abundance Institute, the Information Technology Council, and the R Street Institute.
Adam Therrier, an R Street Institute senior fellow, argued that the AI policy debate often focuses on new regulation while giving less attention to existing laws that already cover, or could come to cover, algorithmic applications. From that view, the sandbox is not just about avoiding new rules; it is also about preventing older rules from slowing AI uses that supporters believe could benefit the public.
Cruz’s one-pager says “most US rules and regulations do not squarely apply to emerging technologies like AI.” Cruz argues that regulators should become more flexible instead of forcing AI developers to design inferior products to comply with outdated Federal rules.
Therrier pointed to potential AI applications in healthcare, transportation, and financial services, saying they “could offer the public important new life-enriching service” unless “archaic rules” block those benefits by standing in the way of marketplace experimentation.
A broader fight over who controls AI regulation
The SANDBOX Act follows Cruz’s earlier efforts to restrict state AI regulation. He previously helped push for a decadelong moratorium on state AI laws as part of Republicans’ “big beautiful” budget bill. More recently, he lost a bid to punish states for regulating AI and ultimately voted against his own measure amid overwhelming bipartisan opposition.
The new framework again targets “burdensome” state AI regulations, as well as foreign ones. After the SANDBOX Act, Cruz expects to introduce more laws to support the framework, likely creating a path for similar future moratoriums that could block state laws.
That makes the bill more than a technical proposal about test environments. It is a statement about where AI oversight should sit, how quickly companies should be allowed to deploy new tools, and how much power agencies should retain when powerful firms ask for exceptions.
The debate now turns on a practical question: whether a faster review system for AI experimentation can protect the public while giving companies room to innovate, or whether it would weaken oversight precisely when critics say stronger scrutiny is needed.