If funded, program would retrain workers hurt by AI or automation

The Labor Secretary would provide approved participants career counseling, retraining, monthly support when not in qualifying work, relocation aid, and wage replacement when eligible. Applicants must be seeking work and either report recent job harm from AI or automation, be recent graduates whose planned work was harmed, or work in a listed at-risk occupation. Applicants must take training or keep seeking work and cannot continue receiving unemployment benefits after acceptance. Annualized monthly support would be capped at the state median wage for recent graduates and 85% of earlier income for other participants. States could choose Medicaid coverage for uninsured participants not yet in qualifying work and their dependents, continuing six months after work begins. Congress could provide $60 billion each fiscal year, but lawmakers would still need to supply the money before it could be spent on the program as described here. The SBA would prioritize participants for covered loans and grants and reserve 1.5% of annual funding for its main business-loan program. Labor officials would also report yearly on technology's job effects, offer online training where feasible, and help participants apply for federal jobs. Most participation and support would last no more than three years. Non-graduates entering qualifying work below their earlier pay could receive declining wage replacement for five years.

Employers other than small businesses would pay quarterly worker-adaptation contributions based on wages and charged program costs, with rates capped at 2%. A saturation period could begin with a single calendar year when the employer's trailing 12-quarter ratio of charged worker-support costs to wages exceeds 2% before the cap. During saturation, declines in U.S. full-time-equivalent positions could trigger a payroll-tax-equivalent charge for up to five years per eliminated position. Restoration at equal or higher pay and benefits must last four quarters, and removing that job within 24 months would restore earlier charges with interest. Affected C corporations would also face graduated surtaxes of 2%, 4%, and 6% for three years. Employers would face tax recapture for faster write-offs claimed on automation assets over the previous five tax years and could not use certain accelerated write-offs for automation assets placed in service during saturation. Program costs would not be charged for specified business sales, bankruptcy terminations, disasters, or similar uncontrollable events. Employer disputes could not delay or reduce worker aid. Eligible nonprofit and government employers could elect to reimburse actual charged costs instead of paying contributions. Contributions would start for calendar years beginning after the first anniversary of enactment. The saturation-period rules would cover periods beginning after that date.

Providers would pay 4%, 6%, and 8% on portions of annual U.S. AI-service receipts above $50 million, $500 million, and $5 billion, respectively. The tax would cover businesses serving U.S. customers with very large AI models, wherever the provider is based. Federal contracts, nonprofits, and colleges are excluded. The rates would be halved to 2%, 3%, and 4% in the first calendar year beginning after enactment. Revenue would enter a worker-adaptation trust fund. Spending laws would have to release the money, first for the worker program and then, if money remains, for child care, elder care, and workforce programs serving people affected by AI or automation. The tax provisions would apply to calendar years beginning after enactment.

Federal agencies would take escalating steps when the national quarterly unemployment rate crosses three thresholds for two straight quarters. Each quarter's rate would be the average of the three seasonally adjusted national monthly unemployment rates. Above 6%, agencies would consider employment effects when reviewing applications for discretionary grants and loans. Above 7.5%, Cabinet agencies would form an emergency task force, the Government Accountability Office would review agencies and recommend job incentives, and the Congressional Budget Office would score employment effects for covered legislation. Above 9%, the Federal Reserve Board would prioritize its maximum-employment mandate in its activities. Each response would stop after unemployment remains below its own threshold for two consecutive quarters.

Employers facing covered plant closings or mass layoffs would give 120 days' notice, while employers could claim a credit equal to 25% of qualifying worker-training costs. The notice rule would cover closings and layoffs governed by federal notice law. The credit would cover credential-based training that moves employees from listed at-risk occupations to other work. Notices must say whether AI or automation substantially caused the closing or layoff and provide information about retraining, unemployment help, and transition resources. Credited training must lead to a recognized college or career credential. Costs for meals, lodging, transportation, and other incidental services would not qualify. The notice amendments would apply to a covered closing or layoff that occurs 180 days after enactment. The credit would apply to tax years beginning after enactment.