The Capitol LedgerEvery move in Congress, on the record
HomeBillsRecordsLedger AI

The Capitol Ledger

Built from the Congressional Record and Congress.gov. Every summary links to its source.

AboutFAQFeedbackTerms of ServicePrivacy Policy
Follow us onlineInstagram

Summaries are AI-generated from primary sources. Verify anything important against the original record.

The Capitol Ledger

Built from the Congressional Record and Congress.gov. Every summary links to its source.

The projectAboutFAQFeedback
LegalTerms of ServicePrivacy Policy
Follow us onlineInstagram

Summaries are AI-generated from primary sources. Verify anything important against the original record.

All bills
S. 2750·119th Congress·Introduced Sep 10, 2025

SANDBOX Act

IntroducedScience, Technology, CommunicationsMajor
View bill text
Sponsor
Sen. Cruz, Ted [R-TX]
Committees
Commerce, Science, and Transportation Committee (primary)
Last action
Sep 10, 2025

Bottom line

The SANDBOX Act aims to accelerate AI innovation by providing a controlled environment for companies to test AI technologies free from certain regulatory burdens, while also establishing a process for Congress to update outdated regulations.

What it actually does

This bill establishes a federal regulatory sandbox program for artificial intelligence (AI) under the Director of the Office of Science and Technology Policy (OSTP). The program allows AI developers to apply for temporary waivers or modifications of existing federal regulations to test and deploy AI products or services on a limited basis, without being subject to full enforcement. It also creates a mechanism for Congress to review and potentially amend or repeal regulations identified through the sandbox program as unnecessary.

Proponents argue

Proponents argue that the bill will foster innovation in artificial intelligence by reducing regulatory barriers that can stifle development and deployment of new technologies. They contend that a regulatory sandbox allows for real-world testing of AI systems in a controlled manner, providing valuable data on their benefits and risks without immediately subjecting them to potentially outdated or overly restrictive regulations. This approach, they say, will help the U.S. maintain its leadership in AI and create new economic opportunities and jobs.

Opponents contend

Opponents might argue that the bill could create regulatory loopholes that expose consumers and the public to undue risks from untested AI technologies, particularly concerning health and safety, economic damage, or unfair trade practices. They may contend that granting waivers could undermine existing consumer protections and regulatory oversight, potentially leading to adverse outcomes that are difficult to reverse. Critics could also raise concerns about the potential for regulatory capture, where industry interests might unduly influence the waiver process or the congressional review of regulations.

The bill is moderately complex due to its procedural nature, requiring careful reading to understand the application, review, and oversight mechanisms. An informed reader would need adequate time to fully grasp its implications.

Section 2, Title VII, Sec. 702(a)

Establishment of the AI Regulatory Sandbox Program

prominently featuredstraightforward

This provision mandates the Director of the Office of Science and Technology Policy (OSTP) to create an Artificial Intelligence Regulatory Sandbox Program within one year of the bill's enactment. This program will allow individuals and entities to apply for temporary waivers or modifications of federal regulations to test, experiment with, or temporarily provide AI products, services, or development methods to consumers on a limited basis, without being fully subject to those regulations.

GroupImpactMechanismScale
GroupArtificial Intelligence Developers and BusinessesImpactBenefitMechanismProvides a pathway to test and deploy AI technologies without immediate full regulatory compliance, potentially accelerating innovation and market entry.ScaleSignificant for companies developing novel AI applications.
GroupFederal Regulatory AgenciesImpactOperational ChangeMechanismRequires agencies to review applications for waivers/modifications and participate in the sandbox program, potentially shifting resources and priorities.ScaleBroad impact across agencies with jurisdiction over AI-related activities.

Supporters argue

Supporters argue that establishing a dedicated sandbox program is crucial for fostering AI innovation by providing a safe space for experimentation, allowing developers to iterate quickly and bring new technologies to market without being hampered by potentially outdated regulations. They believe this will accelerate U.S. leadership in AI.

Critics contend

Critics might contend that creating a sandbox could lead to a 'race to the bottom' in regulatory standards, potentially compromising public safety or consumer protection in the pursuit of innovation. They may argue that existing regulatory frameworks, with appropriate updates, are sufficient.

Tradeoffs

The provision balances the desire to accelerate AI innovation with the need to maintain regulatory oversight and protect consumers. It seeks to reduce barriers to entry for AI technologies while establishing a framework for risk assessment and mitigation.

Section 2, Title VII, Sec. 702(c) and (d)

Application and Agency Review Process for Waivers

prominently featuredmoderately complex

This section details the process for applying for a waiver or modification and how federal agencies will review these applications. Applicants must provide extensive information, including a description of the AI product, the regulations they seek to waive, the potential benefits, and a plan for mitigating risks (health, safety, economic damage, unfair trade practices). Applicable agencies then review these applications, considering public input, and decide whether to grant or deny the waiver, providing a detailed record of their decision.

Section 2, Title VII, Sec. 702(k) and (l)

Terms of Participation and Consumer Protection

prominently featuredstraightforward

This section outlines the conditions for participating in the sandbox program and the consumer protections required. Participants are immune from criminal or civil enforcement of *specifically waived* provisions while in compliance with their agreement, but not for other criminal offenses or existing consumer rights to seek damages. Before deploying AI, participants must publicly disclose their program involvement, the experimental nature of the AI, potential risks, and that they are not immune from civil liability for harm. They must also report any incidents causing harm within 72 hours.

Section 2, Title VII, Sec. 702(n)

Reporting Requirements for Participants and Director

prominently featuredstraightforward

This section mandates regular reporting from both sandbox participants and the Director of OSTP. Participants must submit reports at specific intervals (40 days after commencement, halfway mark, 30 days before expiration) detailing consumer participation, risk assessments, unanticipated risks, adverse incidents, and benefits. The Director must submit an annual report to Congress summarizing program activities, including applications received/approved, benefits/harms, waived provisions, and affected consumers.

Section 2, Title VII, Sec. 703

Congressional Review of Covered Provisions

prominently featuredhighly complex

This section establishes a process for Congress to review and potentially amend or repeal federal regulations based on the experience gained from the AI regulatory sandbox. Annually, the Director of OSTP must submit a special message to Congress detailing which covered provisions (regulations) have been waived or modified, which applications were denied, and which provisions the Director recommends for amendment or repeal, along with a rationale. Congress can then introduce a 'joint resolution of approval' to act on these recommendations, subject to expedited legislative procedures.

FOUND

The definition of 'covered provision' explicitly includes not only formal rules (as defined in 5 U.S.C. 804(3)) but also 'any associated guidance, frequently asked questions publications, bulletins, or associated, derivative material and any rule the adoption of which is expressly required by statute.'

Section 2, Title VII, Sec. 701(6)

Why it matters:This is standard legislative practice to ensure comprehensive applicability of terms. However, it is substantively important as it broadens the scope of what can be waived or modified beyond just formal regulations to include informal agency interpretations and statutory mandates.

Case for: Proponents would argue that including guidance and FAQ documents is essential because agencies often use these informal mechanisms to create de facto regulations that can be just as burdensome as formal rules. Waiving these allows for more comprehensive regulatory relief.

Case against: Critics might argue that waiving informal guidance could create confusion or undermine agency efforts to provide clarity, especially if the underlying statutory mandate remains. Waiving rules expressly required by statute could also raise constitutional concerns about executive overreach or legislative delegation.

Estimated impact: This significantly expands the universe of regulatory requirements that can be temporarily set aside, potentially offering much broader relief to AI developers than if only formal rules were included.

FOUND

If the head of an applicable agency does not submit a record of their decision on an application within the specified deadline (90 days, with a possible 30-day extension), the Director of OSTP 'shall presume that the head of the applicable agency does not object to the granting of the waiver or modification requested by the applicant and may proceed with the application.'

Section 2, Title VII, Sec. 702(d)(5)(C)

Why it matters:This is a common legislative tactic to prevent agency inaction from stalling a program. It shifts the burden of action to the agency, effectively creating an 'approve by default' mechanism if the agency fails to respond.

Case for: Proponents would argue this provision is essential to prevent bureaucratic delays from undermining the sandbox program's goal of rapid innovation. It ensures that agencies are incentivized to act promptly and decisively on applications.

Case against: Critics might argue that this 'default approval' mechanism could lead to waivers being granted without adequate review, especially if agencies are understaffed or overwhelmed. This could inadvertently increase risks to public safety or consumer protection if a critical review is missed.

Estimated impact: This provision could significantly expedite the approval process for some applications, particularly if agencies are slow to respond. It places a strong procedural incentive on agencies to engage with the program.

About this analysis. AI-Generated from the official bill text and available committee reports. Gaps in available data are noted explicitly. Verify important details with the official Congress.gov record.

On this page

  • Executive summary
  • Key provisions
  • Buried treasure
  • Follow the money
  • Critical analysis
  • Questions to ask
  • Implementation
Ask Ledger AI about this