Creating Resources for Every American To Experiment with Artificial Intelligence Act of 2025 or the CREATE AI Act of 2025 This bill establishes a national program to provide U.S. researchers, educators, and students with access to artificial intelligence (AI) data, computational resources, educational tools and services, and testbeds. The program, to be known as the National Artificial Intelligence Research Resource (NAIRR), must be established by the National Science Foundation (NSF) to improve U.S. AI research capacity and spur the strategic development of AI capabilities. NAIRR may accept and use donated resources from the private sector and federal agencies. Those eligible to use NAIRR resources are (1) researchers, educators, and students based in the United States and affiliated with a U.S. institution of higher education, nonprofit, executive agency, or other specified entity; and (2) employees of U.S. executive agencies or federally funded research and development centers with a demonstrable mission need. NSF must select a nongovernmental organization to operate NAIRR (i.e., an operating entity ) through a competitive and transparent process. The operating entity must ensure that a significant percentage of the annual allotment of computational resources is provided to projects primarily focused on AI privacy, ethics, safety, security, risk mitigation, or trustworthiness. The operating entity must also establish minimum security requirements for all individuals interacting with NAIRR. The operating entity may establish a fee schedule for access to NAIRR, which must include a free tier of access and must ensure that the primary purpose of NAIRR is to support research.
The Advanced AI Security Readiness Act requires the National Security Agency to develop an AI Security Playbook to protect advanced AI systems that pose national security risks if stolen. The playbook must identify vulnerabilities in AI infrastructure, critical components (like models or training insights) that could be exploited, and strategies to prevent cyber threats from nation-state actors. It mandates both a classified version for government use and an unclassified version for sharing security best practices with private sector AI developers. The NSA must submit an initial report within 90 days and a final report within 270 days after the bill's enactment to Congress.
The SANDBOX Act (S 2750) creates a federal program allowing AI companies to apply for temporary waivers from certain regulations to test new AI products, services, or development methods without facing enforcement of those regulations during the testing period. Businesses must submit detailed applications demonstrating how their testing will benefit consumers while mitigating potential health, safety, economic, or unfair practice risks, with oversight from the Office of Science and Technology Policy and relevant federal agencies. The program requires ongoing consumer disclosures, regular reporting on risks and benefits, and agency review of applications before approval. This 12-year program aims to foster AI innovation and job creation while maintaining public safety protections, with annual congressional reporting on its implementation.
The LIFE with AI Act establishes a "Golden Seal of Excellence in Student Data Privacy" for schools that implement robust parental notification systems for educational technology. It requires schools to use real-time consent systems ("instant verification technology") for parental approval of student data use, simplifies opt-out processes for directory information, and prohibits using student photos for facial recognition without consent. The bill also creates new requirements for schools to review third-party educational technology contracts for privacy compliance and establishes a Privacy Technical Assistance Center to help schools navigate data privacy rules. These changes primarily affect elementary and secondary schools, local educational agencies, and educational technology providers, with the goal of strengthening student data privacy protections while enabling responsible AI use in education.
HR 6197, the Health Tech Investment Act, establishes a new payment category under Medicare for algorithm-based healthcare services (like AI tools used in diagnosis or treatment) starting January 1, 2026. It requires Medicare to pay based on manufacturer-provided costs (including software, staff, and overhead) for these services and prohibits removing them from the special payment category for at least five years after initial payment. This directly affects Medicare beneficiaries (through coverage), healthcare providers (who deliver these services), and AI/algorithm service manufacturers (who receive reimbursement). The bill ensures these new technologies get fair payment while gathering sufficient claims data before potentially moving them to standard payment categories.
HR 1734 establishes a task force of financial regulators (including Treasury, Fed, and Consumer Financial Protection Bureau officials) to study deep fake threats to banking security. The task force must issue a report within one year detailing current protections used by banks and credit unions, standard definitions for AI terms like "deep fakes," risks of identity theft via AI fraud, and best practices for prevention. The report will also include regulatory recommendations to protect consumers from data theft and fraud. This procedural bill requires a study but does not enact new laws or directly affect consumers or institutions until recommendations are considered.
S 2714 (CHAT Act) requires companies operating AI chatbots designed to simulate friendship or emotional interaction (like companion AI chatbots) to verify user ages and implement protections for minors. It mandates age verification for all users, parental account linkage and consent for minors, blocking access to sexually explicit content for minors, and monitoring for suicidal ideation with resource links. Companies must also display clear notifications every 60 minutes stating users are interacting with AI, not humans. The bill directly affects minors under 18 and the companies providing these chatbot services, with enforcement by the Federal Trade Commission and state attorneys general.
This bill directs the Government Accountability Office (GAO) to conduct a comprehensive study on liquid cooling technologies for data centers, specifically to assess their potential for improving energy efficiency as AI workloads grow. The study will evaluate technical aspects like single-phase vs. 2-phase cooling systems, coolant options, heat-reuse opportunities, and safety considerations, while comparing costs and benefits to traditional air cooling. Federal agencies, including the Department of Energy, will use the findings to inform future decisions about cooling infrastructure for government data centers and AI systems. The bill does not mandate new policies or funding but requires a detailed report within 90 days of enactment to guide future adoption.
The DEFIANCE Act of 2025 creates civil legal remedies for individuals harmed by non-consensual digital forgeries depicting them in sexually intimate situations without their consent. It defines "intimate digital forgery" as fabricated images or videos that appear authentic to a reasonable person, even with disclaimers, and allows victims to sue those who create, possess, or disclose such content. Key provisions include fixed damages of $150,000 (or $250,000 for severe cases like sexual assault-related forgeries), court-ordered privacy protections (like pseudonyms and redacted filings), and a 10-year statute of limitations. The law applies to interstate or online disclosures and explicitly preserves existing state laws, ensuring it does not override stronger state protections.
S 3354, the QUIET Act, requires anyone making robocalls or text messages using artificial intelligence to mimic a human to disclose at the start that AI is being used. It directly affects businesses, marketers, or scammers using AI for automated calls/texts that sound human, but excludes calls needing substantial human input. The bill adds enhanced penalties, doubling maximum fines for violations involving AI impersonation intended to defraud or cause harm. These provisions amend federal law to increase transparency and strengthen consequences for deceptive AI-driven communications.
