At a Joint Economic Committee hearing, witnesses said artificial intelligence offers major economic benefits but that lawmakers should move now to manage workforce disruption and safety risks.

Will Reinhardt, a senior fellow at the American Enterprise Institute, told the panel that generative AI has spread extraordinarily fast and can act as a "general purpose technology" that raises productivity across sectors. "Generative AI can be a skill equalizer," Reinhardt said, citing examples in customer support and software development where tools have boosted output and reduced turnover.

Reinhardt also flagged an uneven early impact on employment: "There is evidence that early career workers aged 22 to 25 have seen employment declines," he said, and warned that a patchwork of state AI rules could impose heavy compliance costs and grant restrictive states veto power over innovation.

Ruth Whitaker, director of technology policy at Third Way, emphasized safety and access. She told the committee that federal engagement to test advanced models has shown how a safety framework might work: "Stress testing" by government and developers gives transparency on safeguards and gaps, she said, and would reassure consumers. Whitaker recommended targeted federal support—such as compute and data access for small firms and apprenticeships—to help smaller businesses compete.

On policy levers, witnesses urged Congress to build on existing law rather than create a large new AI bureaucracy. Reinhardt suggested Congress should "clarify some of the current frameworks" so companies have certainty, while Whitaker pointed to pending legislative tools and standards, including NIST guidance and proposals like the CREATE AI Act, to expand safe adoption and equitable access.

Committee members focused on workforce and competitiveness. Senator Kelly described a proposal to reinvest a fraction of frontier-AI profits into an "AI horizon fund" for training, apprenticeships and community infrastructure; Whitaker said such resources would help small employers bear apprenticeship startup costs. Lawmakers also pressed witnesses for data on which industries and worker cohorts are most exposed; Whitaker said the Department of Labor could do more to track employer adoption and shifting skill demands.

Members raised risks that reach beyond jobs. Reinhardt and others warned about "hallucinations"—incorrect outputs from generative models—and the pollution of models by low‑quality or retracted academic papers, a data-quality problem witnesses said will require better curation and domain-specific training datasets.

The hearing closed with a common theme: witnesses urged a balanced approach that protects safety and fairness without stifling the innovation they said can deliver medical advances, new industries and productivity gains. Senators asked for written follow-ups, which the panel agreed to provide. The committee adjourned with no formal votes taken.