S. 3339 — AI Workforce PREPARE Act

This section directs the Secretary of Labor, not later than 45 days after enactment, to issue a request for comment on a publicly accessible DOL website allowing interested persons a 60-day period to submit views on implementing the Act, including (1) data collection, forecasting tools, and data-sharing efforts; (2) reforms to increase researcher access to Census Bureau, Bureau of Labor Statistics, and other federal data; (3) AI workforce forecasting best practices; (4) improving training grant programs; (5) evaluating information value; (6) data governance and security; and (7) other topics determined appropriate by the Secretary. Not later than 120 days after enactment, the Secretary must post received comments on the website. This section further requires the Secretary, not later than 180 days after enactment, to convene an initial workshop with invited economic and technical experts on AI (e.g., economists, academics, industry and labor representatives, forecasting practitioners), agency representatives (e.g., Census Bureau, OSTP, CEA), labor organizations, grant entities, the Workforce Information Advisory Council, GAO, and other stakeholders representing diverse viewpoints to discuss comments, implementation topics, critically evaluate BLS AI workforce impact assumptions, and rank the value of proposed data and policy efforts.

This section authorizes the Secretary of Labor to appoint up to 20 covered individuals—experts in artificial intelligence, machine learning, advanced data science, or related fields—to excepted service positions in the Department of Labor without regard to competitive service hiring requirements (5 U.S.C. ch. 33), position classification (5 U.S.C. ch. 51), or General Schedule pay banding (5 U.S.C. ch. 53, subch. III). Appointments are for up to 24 months, extendable by one additional 24-month period upon certification to Congress and the President of the individual's continued value; the appointment authority terminates 4 years after enactment (incumbents may continue serving). The Secretary may set basic pay up to the GS-15, step 10 rate; provide recruitment, relocation, retention incentives; and performance awards without regard to title 5, U.S. Code, subject to an aggregate annual compensation cap equal to the Vice President's total annual pay under 3 U.S.C. 104 (i.e., basic pay plus locality pay, as adjusted annually per General Schedule increases and the Employment Cost Index). Appointees primarily perform duties related to measuring and forecasting artificial intelligence's labor market impacts and policymaking for workforce preparation or adjustment; they may be deemed employees for specified federal benefits (e.g., leave, retirement). The Secretary must submit annual reports to the Senate Committee on Health, Education, Labor, and Pensions and the House Committee on Education and the Workforce detailing appointee numbers, qualifications, duties, and impacts. Authorizes $6 million for FY2026 through FY2030.

This section establishes an Artificial Intelligence Workforce Research Hub in the Department of Labor, to be set up not later than 90 days after enactment, to conduct research in collaboration with the Bureau of the Census, Bureau of Economic Analysis (Department of Commerce), and Bureau of Labor Statistics evaluating artificial intelligence's impact on the U.S. workforce—including recurring analyses, scenario planning for potential AI impact levels, and actionable insights for training programs, unemployment mitigation, and other employment effects. The section authorizes details to the Hub, with or without reimbursement, of federal employees (including those appointed under section 102), private sector employees (under 5 U.S.C. §§3701-3704), and state or local government employees (under 5 U.S.C. §§3371-3376), without interruption of civil service status or privileges. The Secretary may fund the Hub using existing amounts available for training program research or employment impact research—including funds for the Employment and Training Administration or Bureau of Labor Statistics—with no additional appropriations authorized; the Hub's requirements sunset 4 years after enactment.

This section establishes a four-year pilot project at the Census Bureau, in consultation with the Secretary of Labor and Chief Statistician of the United States, to produce a recurring statistical series on job-to-job occupational flows—similar to the Census Bureau's J2J Explorer tool but at a more detailed Standard Occupational Classification (SOC) code level—for at least 15 AI-impacted occupations designated by the Secretary of Labor not later than 240 days after enactment and every two years thereafter. The Secretary of Labor's designations, published in the Federal Register, consider factors such as potential AI-driven worker dislocation or shortages, analytical value, data collection feasibility, and data gaps. The Census Bureau Director must publish the first series for one quarter not later than 18 months after enactment and at least one quarter per fiscal year thereafter, using sources such as federal surveys, administrative records, private payroll data partnerships, and state/local pilots; if production proves impracticable or too costly, the Director must instead submit a public report on barriers, recommended changes, and a cost-benefit assessment. This section further directs the Commissioner of Labor Statistics, in consultation with the Chief Statistician and Census Bureau Director, to assess proposals for secure remote researcher access to related individual-level data and publish a report on their value, risk, and cost not later than one year after enactment, potentially leveraging the National Secure Data Service demonstration project (42 U.S.C. 19085). This section prohibits additional appropriations to carry out these activities.

This section requires the Secretary of Commerce, acting through the Director of the National Institute of Standards and Technology (NIST), to commence not later than 270 days after enactment at least one prize competition under the federal prize competition authority (15 U.S.C. 3719) to develop benchmarks or similar reproducible methods quantitatively measuring artificial intelligence's ability to automate or augment tasks or occupations, primarily to improve forecasts of artificial intelligence impacts on workers and retraining needs. (As background, the federal prize competition authority allows agencies to award prizes to stimulate innovation advancing agency missions.) In carrying out the competition, NIST shall consult with the Secretaries of Commerce and Labor, the Commissioner of the Bureau of Labor Statistics, and the Director of the National Science Foundation. NIST shall develop the competition's structure and evaluation criteria to further that purpose, which may include (1) categories by occupation impacted or artificial intelligence capability domain—selected considering labor market informativeness (e.g., number of workers affected and their economic security), submission quality, complementarity to existing benchmarks, private underfunding relative to public value, and other factors; (2) phases such as design and prototype; and (3) criteria mitigating common benchmark issues (e.g., undetected training data contamination, rapid loss of discriminatory value from performance ceilings). NIST may award companion grants or cooperative agreements to support benchmark design, construction, validation, and maintenance—including data collection and labeling, evaluator training and rubrics, and third-party replication and inter-rater reliability testing—and may contract with for-profit or nonprofit entities or state, territorial, local, or Tribal agencies to administer the competition and related activities. This section authorizes $7 million to the Secretary of Commerce for FY2026 through FY2030.

This section establishes a voluntary program, administered by the Secretary of Labor through the Commissioner of Labor Statistics in coordination with the Secretary of Commerce and the Director of the Office of Science and Technology Policy, to enable artificial intelligence (AI) developers, AI deployers, and other private entities to share anonymized data on AI adoption and use in the workforce. The Secretary may enter into memoranda of understanding or other data-sharing agreements; data must be used exclusively for statistical purposes as defined in 44 U.S.C. 3561 (i.e., to develop or produce evidence from identifiable data through statistical activities for statistical purposes), treated as confidential under 44 U.S.C. ch. 35 subch. III, exempt from disclosure under 5 U.S.C. 552(b)(3), and accompanied by descriptions of sources, methodology, and limitations; and data may not be used for regulatory enforcement or antitrust actions. The section further requires development of statistical methodologies for data validity, representativeness across workforce sectors and demographics, quality standards, and bias mitigation; maintenance of a public roster of participating entities on the Department of Labor website; publication of machine-readable aggregate statistics at least semiannually (unless deemed detrimental); and a report to Congress within two years evaluating participation, data quality, usefulness for labor-market analysis, barriers, and recommendations for program changes. It authorizes appropriations of $7 million for FY2026 through FY2030 and defines AI deployer (an entity operating AI for internal or third-party use, including cloud providers) and AI developer (an entity that builds, designs, codes, produces, trains, or owns AI, excluding pure deployers).

This section directs the Secretary of Commerce, acting through the Director of the Bureau of the Census, and the Secretary of Labor, acting through the Commissioner of the Bureau of Labor Statistics—not later than one year after enactment and in consultation with the Chief Statistician of the United States—to revise specified surveys to incorporate or improve questions on the adoption and use of artificial intelligence (AI). The surveys are (1) for the Census Bureau: the Annual Business Survey, Current Population Survey, and Business Trends and Outlook Survey; and (2) for the Bureau of Labor Statistics: the Occupational Requirements Survey and American Time Use Survey. The questions must, to the extent practicable, be informed by information collected under section 101 and other appropriate sources and seek data on (A) types of AI adopted (e.g., language models vs. simple predictive models); (B) occupations or tasks most impacted; (C) changes in skill requirements or training investments due to AI, including skill shortages for AI development, operation, or integration; (D) intensity and frequency of AI use; and (E) work outcomes attributable to AI (e.g., changes in time spent or output quality or quantity). The Secretaries may narrow the scope of AI covered (e.g., limiting to language models or excluding non-frontier AI forms), notwithstanding the Act's general definition, and must consult to promote consistency where feasible. No additional amounts are authorized to be appropriated.

This section establishes disclosure requirements in Worker Adjustment and Retraining Notification Act (WARN Act) notices for mass layoffs (i.e., employment losses at a single site affecting either 500 employees or one-third of the active workforce if at least 100 employees) where artificial intelligence (AI) is a substantial factor, as determined under Department of Labor (DOL) guidance issued not later than 300 days after enactment. (As background, the WARN Act generally requires employers with 100 or more employees to provide 60 days' advance written notice of a plant closing or mass layoff to affected employees or their representatives, state dislocated worker units, and local government officials.) Such notices must include (1) a statement specifying that AI was a substantial factor; (2) the type and usage of such AI; (3) an estimate of the percentage of employment losses attributable to AI; and (4) actions, if any, taken to upskill or retrain employees prior to the layoff. Employers may satisfy these requirements through good-faith statements, including estimates; the provision applies to notices issued on or after one year after enactment of the AI Workforce PREPARE Act.

This section requires the Secretary of Labor to publish in the Federal Register, not later than 240 days after enactment and every two years thereafter, a list of at least 15 occupations at the six-digit Standard Occupational Classification (SOC) code level deemed highly valuable for deeper analysis of artificial intelligence (AI) impacts, considering factors such as expected changes in employment demand from AI, current employment size, policymaking relevance (e.g., for Federal workforce or education grants), analytical feasibility, and others; the Secretary must consult the Workforce Information Advisory Council and the Office of Science and Technology Policy and consider public input. Not later than six months after the first list and annually thereafter, the Secretary must publish a report with prediction interval forecasts (20th to 80th percentile range or Secretary-approved alternative) of employment levels for listed occupations over 2-, 4-, and 8-year periods, reflecting holistic uncertainty from technological changes including AI but excluding acute shocks (e.g., recessions, wars, pandemics); the report must include benchmark forecasts (e.g., trend-extrapolation models or generative AI), methodological details and data sources for replication, gap assessments, and supplementary analyses. The Secretary must evaluate each forecast's accuracy, including quantitative assessment, not later than 120 days after relevant official employment data are released.

This section establishes a recurring prize competition administered by the Director of the National Science Foundation (NSF) to incentivize accurate forecasts and rationales for short-horizon questions on labor-market implications of artificial intelligence. Eligible U.S. citizens, permanent residents, or U.S.-incorporated entities (excluding federal entities and employees) may compete on topics including AI model performance benchmarks, adoption indicators, occupation-level employment changes, and other metrics deemed valuable in consultation with the Secretary of Labor; competitions occur and are scored at least every six months using criteria of forecasting accuracy and rationale informativeness. The NSF Director must consult the Secretary of Labor, Director of the Office of Science and Technology Policy, and forecasting organizations; post annual public summaries of results to inform federal labor-market forecasts or policy; retain data for at least five years after sunset; and, by three years after enactment, publish a report evaluating program effectiveness and recommending continuation or modification. The section authorizes $6 million over FY2026 through FY2030 and sunsets the competition five years after enactment (with obligated prize funds distributable thereafter).

This section directs the Secretary of Labor to submit a report to the Senate Committee on Health, Education, Labor, and Pensions and the House Committee on Education and the Workforce, not later than two years after enactment, describing the incorporation of data, benchmarks, and forecasts developed under this Act into recipient selection and performance measurement criteria for (1) formula and competitive grants under title I of the Workforce Innovation and Opportunity Act (WIOA), (2) apprenticeship and pre-apprenticeship programs and related training grants administered by the Secretary, and (3) other appropriate Department of Labor grant programs. The report must identify statutory or regulatory barriers to such use, describe support for states, localities, and applicants in utilizing the data to improve applications and performance, and provide recommendations on timelines, responsible offices, and resources; it requires consultation with the Workforce Information Advisory Council. The Secretary must publish the report on the Department of Labor website in machine-readable format within 30 days of submission and authorizes $500,000 for FY2026 through 2030.

This section directs the Secretary of Labor, not later than 12 months after enactment, to conduct (directly or by grant) a study on potential design options and challenges for a Rapid Artificial Intelligence Adjustment Assistance Program to aid workers dislocated or job-seekers impacted by artificial intelligence. The study must draw on data and forecasts from this Act and lessons from the Trade Adjustment Assistance program (i.e., aid for workers displaced by imports under 19 U.S.C. 2271 et seq.) and other adjustment programs; examine specified questions on worker eligibility determination (including data needs, regulatory barriers, AI use in adjudication, petition processes), assistance forms and effects, legal compatibility, performance evaluation, costs, collaboration, and other topics; and be completed and published on the Department of Labor website not later than two years after enactment. This section authorizes $1 million for FY2026 through 2030 to carry out the study.

This section requires each state and local workforce development board receiving funds under subtitle B of title I of the Workforce Innovation and Opportunity Act (WIOA) to consider information or forecasts produced under title III of this Act when updating its lists of in-demand industry sectors and occupations—as defined in WIOA section 3 (29 U.S.C. 3102), which are used to prioritize workforce services, training, and resource allocation—beginning not later than one year after enactment and for the subsequent four years. The section further directs the Secretary of Labor to provide states with technical assistance, including compilations of resources and data harmonization guidance, and authorizes $3 million to carry out this section for FY2026 through 2030.

This section directs the Secretary of Labor to (1) identify, in consultation with states, local workforce development boards, private employers, and other entities, data elements relevant to data collection and reporting under this Act pertaining to the workforce, job vacancies, hiring, earnings, education, skills, and related implementation aspects; (2) submit a report to the Senate Committee on Health, Education, Labor, and Pensions and the House Committee on Education and Workforce within 12 months of enactment detailing how such data could be reported in a standardized manner, collected in a secure repository, and made accessible to researchers; and (3) lead a voluntary, consensus-driven effort with the National Science Foundation and other federal agencies to develop federal standards—and facilitate state and local standards—for producing trusted artificial intelligence-related data under this Act and to support their consistent implementation. The section authorizes appropriations of $3 million for FY2026 through 2030.