Key Takeaways
- The Strengthening Artificial Intelligence Normalization and Diffusion by Oversight and Experiment (SANDBOX) Act allows AI companies to avoid federal restrictions.
- Opponents argue that the SANDBOX Act gives excess federal power, allowing potentially unsafe or risky behaviors that restrictions are meant to prevent.
- Proponents cite the potential for U.S. innovation and significant steps forward in the international market.
What is the SANDBOX Act?
On September 10, 2025, U.S. Senator Ted Cruz (R-TX) introduced the Strengthening Artificial Intelligence Normalization and Diffusion by Oversight and Experiment (SANDBOX) Act. The bill establishes a mechanism for artificial intelligence (AI) developers and companies to apply to the Office of Science and Technology Policy (OSTP) director for a waiver or modification to rules or regulations under any federal agency. It is part of Cruz’s five-pillar framework for American leadership in AI, and relates specifically to the first pillar: unleashing American innovation and promoting long-term growth. While supporters of the bill emphasize its role in bolstering technological innovation, its opponents are growing increasingly concerned that if the bill passes, the rights and privacy of American consumers would be sacrificed in the name of technology.
Where Did the SANDBOX Act Come from?
The bill, originally introduced by Senator Cruz, creates a regulatory “sandbox”—a policy endorsed by President Donald Trump’s AI Action Plan—which gives companies developing AI technologies the freedom to test and launch new technologies without the restrictions of federal rules and regulations. The bill, according to Cruz, prevents companies from being restrained by “outdated or inflexible federal rules” and increases U.S. ability to keep up with China’s evolving technological market. Cruz stresses that following this new framework for AI will “turbocharge economic activity” and “cut through bureaucratic red tape”. The current director of the OSTP, Michael Kratsios, supports Cruz’s claims, referring to current regulations on AI as “anti-innovation” and “a huge problem for our industry”.
If passed by Congress, the bill would allow agencies that oversee federal rules to consider companies’ applications for regulation exemptions. These could last for up to two years at a time. It would also require companies to outline the potential safety and financial risks associated with waiving these regulations and explain how they would mitigate these risks.
What Does the SANDBOX Act Do?
If a company that develops AI applications is currently seeking to apply for a waiver under the new SANDBOX Act, an application would be submitted to the OSTP director. The OSTP director would then forward it to the agency responsible for the provisions your company seeks to have waived. At this point, the agencies contacted have 90 days to review the application, with a possible one-time 30-day extension. If the head of the agency does not respond within the allotted time, the OSTP director is legally allowed to presume that the agency has approved the application.
It is worth noting that similar bills have been introduced in the past. For instance, Arizona was the first state to sign a sandbox into law for financial technology companies, or fintechs, in 2018, allowing startups, entrepreneurs, and even established companies to launch products on a “limited, temporary scale to consumers to test innovative products, services, business models, and delivery mechanisms.” Likewise, Ohio introduced a regulatory sandbox for novel financial products in 2022, allowing products that incorporate new or emerging technologies to “address a problem, provide a benefit, or otherwise offer a product, service, business model, or delivery mechanism” that is not widely available in the state.
Both examples took place at the state level. They applied only to specific industries, such as financial technology, allowing companies to test their products under limited waivers lasting one to two years. In contrast, Senator Cruz’s proposal would operate at the federal level and allow companies from any industry using AI to request waivers. Waivers would initially last two years, but could be extended for up to ten years.
Who Opposes the SANDBOX Act?
The SANDBOX Act is effectively a mechanism for cutting through bureaucratic red tape, which raises questions of safety: essentially, if the red tape exists for a reason. According to opponents of the bill, allowing companies to be approved solely based on the lack of response from the agencies is incredibly risky, given the high number of requests that the OSTP will receive from companies aiming to start testing and launching their AI technologies.
Further, critics take issue with the bill’s lack of limits on the number of regulations that businesses can have waived, fast tracking the potential for companies to use this opportunity as a “fast pass around regulations.” When a company applies to waive a federal law, they choose the “covered provision” (i.e., an agency’s law) they want waived. Because there is no limit on the number of provisions that can be included in an application or the number of applications a company can file, a company may simply apply to have all restrictions lifted. The agencies can control how many of those applications are approved or denied, but opponents note a risk in the fact that companies are allowed to make a case for waiver of all regulations.
Additionally, the term “covered provision,” as outlined in the federal code, covers nearly all agency regulatory actions with three significant exceptions: first, regulations that govern accounting standards, compensation, and firm organization; second, rules about the management or personnel of an agency; and third, rules that do not substantially affect the rights or obligations of organizations outside of the federal government. Ultimately, under this definition, everything “from consumer safety rules, to water quality standards, to workplace safety requirements” is fair game, giving the executive branch of the U.S. government the power to do as they see fit, regardless of if it puts consumers’ rights and safety at risk.
Moreover, opponents argue that this bill will put customers in danger by quickly advancing new technologies. Companies that build untested, unsafe AI tools could get “hall passes” from the very rules designed to protect customers. According to Public Citizen, the SANDBOX Act “guts basic consumer protections” by allowing companies to evade accountability measures and treat American citizens as test subjects for unregulated technologies.
Who Supports the SANDBOX Act?
Those in favor of the bill, on the other hand, highlight that sandboxes allow for a great deal of innovation, as they authorize companies to explore technologies within a controlled framework, gaining access to testing and “real-world refinement needed for innovation.” Instead of restricting companies to current regulations, the SANDBOX Act attempts to simultaneously allow companies to explain how the current provisions affect them and allow the government to observe the effectiveness of the technology regulations it currently upholds.
This increased innovation could be crucial if American-developed AI companies aim to become competitive with other nations. Allowing companies to explore their technologies in a controlled environment is essential to advancing the types of products that can actively compete with other countries in the international market. In this vein, the U.S. is focused on China as well as other countries. For instance, the United Kingdom has created its own AI Growth Lab, a self-titled “cross-economy sandbox,” which functions similarly to the sandbox proposed by Cruz. It allows regulatory modifications for companies to develop their AI technologies without an “excessive or bureaucratic regulatory process.”
The benefits of adopting regulatory sandboxes are evident internationally. According to a 2020 World Bank analysis of international sandbox programs, 88 percent of governments report that sandboxes have successfully attracted new businesses and experimentation to their financial markets, while 55 percent believe sandboxes have helped foster market competition with fewer regulatory barriers for newly founded and under-resourced companies.
Conclusion
If passed, the SANDBOX Act will test the balance between innovation and consumer protection. While some oppose deregulation in emerging industries such as AI, others argue that regulatory sandboxes could make domestic AI development relevant on a global scale. The outcome of this legislation will signal the U.S.’ priorities related to AI development, consumer protection regulations, and competition in global technology markets moving forward.
FAQ
Q: What is a sandbox?
A: According to Harvard’s Information Technology department, a sandbox is a “secure environment in which to explore Generative AI, mitigating many security and privacy risks.”
Q: Why does the SANDBOX Act matter to U.S. Senator Ted Cruz (R-TX) right now?
A: AI is becoming increasingly important, and the U.S. may be in competition with other countries, such as China, in development and innovation related to AI.
Q: Why is President Donald Trump’s AI Action Plan relevant?
A: It places the advancement of AI at the forefront, and notes an effort to reduce red tape and regulations.