U.S. Senator Ted Cruz (R-TX) speaks during a press conference at the U.S. Embassy, in Mexico City, Mexico, Aug. 29, 2025.
Eloisa Sanchez | Reuters
A new bill from Senate Commerce Chair Ted Cruz, R-Texas, would take a step toward regulating artificial intelligence when it comes to the ability for companies to develop new services.
The proposal would require the federal government to create a contained and secure testing area for AI software, known as a regulatory sandbox. Companies would be able to apply to join through the White House Office of Science and Technology Policy.
Cruz’s effort underscores a concern among many lawmakers that the U.S. will fall behind in AI, particularly to China, if regulations aren’t loosened to allow for more and faster innovation. With the sandboxing proposal, companies could identify regulations that are obstructing their efforts, and ask for a waiver.
For example, a company developing cancer-screening software could show why it needs modifications to Health Insurance Portability and Accountability Act, or HIPAA, regulations in order to test its product. The company could work with the government to determine if a waiver could be granted without violating standards such as patient privacy, safety and consumer protection.
The government would be able to grant waivers for two-year periods for up to 10 years. After 12 years the program would sunset.
Cruz is positioning AI as a top priority in the committee, with a focus on enshrining parts of the Trump administration’s AI Action Plan into law. The committee is scheduled to hold a hearing Wednesday with Michael Kratsios, director of the White House OSTP.
Cruz said in a statement to CNBC that the executive order alone won’t be sufficient to keep the U.S. as a leader in AI development and deployment, and that the bill would help spur innovation.
“Following this new AI framework can turbocharge economic activity, cut through bureaucratic red tape, and empower American AI developers while protecting human flourishing,” he said.
According to a five-pillar plan shared with CNBC, Cruz also aims to create federal standards for AI, prevent AI from being used in harmful ways such as scams, allow free speech and address ethical concerns.
Sandboxes are already in use in other countries such as Singapore, Brazil and France. While Cruz is still working to get Democratic sponsors for his bill, the issue has support from across the aisle. A bipartisan group of lawmakers introduced a separate regulatory sandboxing bill focused on AI use in the financial sector.
