The AI Tax Integrity Act of 2026 directs the Treasury Department to launch a pilot program using artificial intelligence to detect tax fraud, identity theft, and errors in returns prepared by third parties. This initiative is designed to target individual and business taxpayers who may file inaccurate returns, with the program running for a minimum of 18 months and a maximum of two years. Upon completion, the Comptroller General must submit a report detailing the amount of improper refunds recovered, the total government recovery, and the accuracy of the AI tools used during the pilot.
This bill establishes a pilot program under the National Institute of Standards and Technology to test voluntary methods for disclosing when content is generated or manipulated by artificial intelligence. The program requires the NIST Director to work with private companies, civil society, and academia to evaluate how users can be informed about AI interactions and to develop guidelines for such disclosures. Upon completion, the NIST Director must submit a report to Congress detailing the program's findings and offering recommendations for future action. The legislation defines key terms like artificial intelligence and content provenance to ensure clarity in the pilot's scope. Ultimately, the bill aims to explore voluntary industry standards rather than mandating immediate legal requirements for AI labeling.
The AI Security and Innovation Act establishes a new Center for AI Security and Innovation within the National Institute of Standards and Technology to assess risks and improve the security of artificial intelligence systems. This center will evaluate threats such as data leaks and model tampering, develop voluntary safety standards, and conduct research on advanced AI capabilities developed by both U.S. and foreign entities. The bill defines key terms like "artificial intelligence model" and "covered frontier system," and authorizes $20 million annually from 2027 to 2032 to fund the center's activities. While the center director will consult with various federal agencies and AI developers, the legislation explicitly prohibits the center from having regulatory or enforcement powers. The center is authorized to operate for five years before its provisions expire.
The Responsible Artificial Intelligence Defense Act of 2026 establishes a new framework for the U.S. Department of Defense to develop and deploy autonomous weapon systems and artificial intelligence capabilities. It requires the Secretary of Defense to maintain continuous human oversight, ensuring that operators can intervene to stop systems acting illegally or unexpectedly. The bill mandates rigorous testing, cybersecurity reviews, and legal assessments before these technologies are prototyped or fielded, while also creating a working group to advise on safety standards. Specific prohibitions prevent the use of AI for nuclear launch decisions or warrantless surveillance of individuals within the United States. Additionally, the act defines different risk levels for these systems and requires annual reports to Congress on their administration.
The Prior Authorization Accountability Act requires health insurance plans and issuers to publicly report detailed data on their prior authorization processes starting in 2027. This includes submitting information on approval and denial rates, appeal outcomes, processing times, and the specific use of artificial intelligence or other automated technologies in making coverage decisions. The bill mandates that these reports be available on public websites for both individual and group coverage, allowing consumers to compare how different plans handle requests for medical services. Additionally, the legislation updates the Affordable Care Act to ensure that health plans sold through insurance exchanges display this new transparency data to help shoppers make informed choices.
The Bureau of Industry and Security License Administration Enhancement Act aims to streamline and clarify how export control licenses are managed by the U.S. government. It requires that specific regulatory guidance letters be published in official government records within 60 days unless they are converted into permanent regulations, ensuring greater transparency for businesses. Additionally, the bill mandates the creation of specialized advisory committees focused on critical technologies like artificial intelligence, semiconductors, and biotechnology to advise the Secretary of Commerce on export control policies. These committees will consist of industry, national security, and academic experts who must meet regularly and submit annual reports on technological developments and policy recommendations. Finally, the legislation directs a review of existing rules regarding advanced computing integrated circuits, with a report to Congress on the findings and any necessary updates within 120 days of enactment.
The SAFE KIDS Act requires companies providing AI chatbots to verify user ages, conduct regular safety risk assessments, and implement specific safeguards to protect children from harms such as self-harm, sexual exploitation, and manipulative design. Providers must offer robust parental controls that allow parents to limit usage, restrict data collection, and receive notifications if their child encounters dangerous content or expresses suicidal ideation. The legislation also prohibits targeting children with advertisements or selling their personal information without explicit parental consent, while mandating annual independent audits to ensure compliance. Enforcement of these rules will be handled by the Federal Trade Commission, which has the authority to impose civil penalties for violations.
The Web of Biological Data Act of 2026 directs the Department of Energy to create a centralized online platform called the Web of Biological Data to help researchers access and analyze biological information using advanced tools like artificial intelligence. This system will serve as a single entry point for federally funded data, featuring strict cybersecurity measures and tiered access controls to protect sensitive information. The bill establishes a phased rollout, beginning with a pilot version within two years that focuses on specific data types and user-friendly design, followed by a full expansion five years later that includes broader data compatibility and bioinformatic tools. To ensure the project's success, the legislation creates an advisory board with members from industry, academia, and government, and mandates regular reports to Congress on progress and security assessments. Funding is authorized to support the initial development, testing, and long-term maintenance of the platform while preserving existing privacy and ethical protections for biological data.
This bill directs the Department of Homeland Security to conduct annual assessments for five years on how foreign terrorist organizations use generative artificial intelligence to spread extremist messages or develop weapons. These reports must analyze specific incidents from the previous year and include recommendations for countermeasures while protecting privacy and civil liberties. The unclassified portions of these assessments will be made public on a government website, and federal agencies will share relevant intelligence with state and local fusion centers to improve information sharing.
This bill creates a legal pathway for citizens to sue federal employees if they force artificial intelligence companies to censor content based on political views, religious beliefs, or other personal ideologies. It specifically prohibits government workers from directing AI providers to remove specific outputs, alter how the technology is trained, or deny access to services because of a user's viewpoint or identity. The law also bans federal officials from spying on or penalizing individuals for their lawful use of AI tools. While it allows victims to seek damages and legal fees, it does not permit lawsuits against the federal government itself for actions taken within the scope of official duties.
