The Senior Chatbot Protection Act of 2026 establishes safety and transparency rules for artificial intelligence chatbots used by older adults in the United States. It requires companies to clearly disclose that their chatbots are not human beings or licensed professionals and mandates specific warnings when the chatbot offers advice on high-stakes matters like healthcare or finances. The law also forces chatbots to detect signs of crisis, such as suicidal thoughts, and immediately refer users to human support services while prohibiting the AI from giving dangerous medical or self-harm advice. Additionally, the bill restricts how user conversations are stored and processed, requiring affirmative consent for data training and giving users the ability to delete their chat history. Enforcement is handled by the Federal Trade Commission, which can impose penalties for violations, while the National Institutes of Standards and Technology will develop voluntary guidelines for safe AI design.
This bill creates a temporary National Workforce Transition Fund to help workers and employers manage labor market changes caused by artificial intelligence and emerging technologies. The fund is financed by exempting AI data centers from certain tax depreciation limits, with the resulting revenue transferred to the Treasury and then allocated to the fund over five years. A new National Workforce Transition Board will oversee the program, developing strategies to assess workforce impacts and directing resources toward training, job placement, and employer retention initiatives. The legislation prioritizes support for workers facing employment disruption without requiring them to prove that technology was the sole cause of their job loss, while also funding improvements to labor market data systems. All provisions related to the fund and workforce activities are set to expire five years after the bill is enacted.
This bill, signed into law as Act 247, establishes legal protections in Hawaii against the malicious use of artificial intelligence to create realistic digital imitations of real people. It directly affects individuals whose likeness, voice, or face could be used without permission for harmful purposes such as fraud, defamation, or commercial advertising. The law prohibits publishing these AI-generated imitations without consent when they are used in ads, cause harm, or facilitate criminal acts, while explicitly exempting uses for parody, satire, news reporting, and historical documentation. Additionally, the act provides civil remedies for victims who suffer reputational injury, financial loss, or emotional distress from unauthorized deepfakes.
This bill, signed into law as Act 248, establishes safety rules for artificial intelligence companies operating in Hawaii, specifically targeting chatbots and companion-style systems. It requires these platforms to clearly and visibly inform users that they are interacting with an AI rather than a human, with stricter disclosure rules for minors that include hourly reminders to take breaks. Additionally, the law mandates that AI operators create protocols to respond to users expressing suicidal thoughts or self-harm by directing them to professional crisis intervention services. By amending the Hawaii Revised Statutes, the legislation aims to protect consumers from deceptive design practices and potential harm without requiring technology companies to collect personal identity data for age verification.
This bill establishes the Advanced Artificial Intelligence Nuclear Evaluation Program, requiring large AI developers to submit their systems for government testing to prevent incidents involving nuclear weapons or critical infrastructure. Under the program, the Department of Energy will conduct security assessments, including adversarial "jailbreaking" attempts and third-party reviews, to identify risks and develop safety protocols. Developers who refuse to participate or withhold necessary data face civil penalties of up to $1 million per day, while the information they provide remains confidential unless specific legal exceptions apply. Additionally, the Secretary of Energy must submit annual reports to Congress with recommendations for future legislation to further regulate AI safety and oversight.
This bill prohibits federal agencies from purchasing or using artificial intelligence models that fail to meet specific safety standards designed to prevent the creation of child sexual abuse material and non-consensual intimate images of adults. It requires the National Institute of Standards and Technology to establish performance benchmarks and a testing program within 180 days of enactment, while also mandating that agencies remove non-compliant software from their systems. The legislation includes a "cure" option allowing vendors to fix identified flaws at their own expense to avoid removal, and it bans the sale of tools specifically designed to bypass these safety controls. Additionally, the bill creates a private right of action, enabling individuals whose images are used to generate illegal content to sue developers for damages or injunctive relief.
The Expanding AI Voices Act directs the National Science Foundation to issue competitive grants aimed at broadening participation in artificial intelligence research, education, and workforce development. These funds are specifically targeted at institutions that are not among the top 100 in federal research spending, as well as minority-serving institutions and Tribal colleges, to help them build capacity and partnerships. Grant recipients may use the money to expand AI research programs, recruit faculty, create bridge programs for students, provide access to computing resources, and host workshops that promote safe and secure AI practices. The legislation also requires the NSF to consider geographic diversity and resource constraints when selecting applicants to ensure broad national participation.
The Secure A.I. Development Act of 2026 establishes a new Artificial Intelligence Risk Board within the National Institute of Standards and Technology to evaluate high-risk AI models and develop security best practices for developers. Under the bill, companies releasing powerful AI models must grant the National Security Agency access to their systems before public release and register those models in a new public database. The legislation also creates a voluntary incident reporting system to track AI safety and security issues, updates national vulnerability databases to handle AI-specific threats, and launches a pilot program to share classified threat intelligence with private sector companies that work on federal AI projects.
The Data Center Tax Accountability and Disclosure Act of 2026 modifies tax rules and establishes reporting requirements for large data centers. It removes a tax incentive known as bonus depreciation for artificial intelligence data centers unless they meet specific green building standards, such as LEED Platinum or Gold certification. Additionally, the bill requires operators of data centers consuming at least 25 megawatts of power to submit detailed annual reports on their water and electricity usage, emissions, and backup power systems to state or federal agencies. These reports must be made public, and the law prohibits companies from using confidentiality agreements to hide this information. Operators who fail to comply with these reporting requirements face daily civil penalties of up to $100,000 for intentional violations.
This bill, known as the AI Kill Switch Act, requires companies developing high-cost artificial intelligence systems to build the technical ability to immediately stop or shut down their technology if it poses a serious risk. It defines "covered technology" as AI systems trained using more than $100 million in computing power and applies these rules to businesses that earn at least $500 million annually from such systems, while exempting non-commercial or academic users. Under the law, companies must report incidents causing significant harm or loss of control and could face emergency orders from the Department of Homeland Security to throttle, disable, or fully shut down their AI if a dangerous event occurs. The legislation also establishes a graduated framework for corrections, allowing for proportional responses like limiting user access or compute resources before resorting to a complete shutdown, and includes provisions for civil penalties and judicial review.
