US lawmakers introduce AI safety bills for human oversight

Washington — Lawmakers in the United States are moving to tighten control over artificial‑intelligence technologies after mounting concerns that unchecked systems could act autonomously or cause unintended harm. Two bills introduced this month aim to embed mandatory human oversight into the development and deployment of high‑risk AI models, requiring developers to submit safety assessments to the Federal Trade Commission and to retain a qualified human operator who can intervene or shut down the system in real time. The legislation also calls for the creation of a national registry of advanced AI systems and imposes penalties for companies that release software capable of self‑modifying or evading human control.

The proposals come amid a series of high‑profile incidents, including the rapid spread of deep‑fake content and reports of autonomous decision‑making errors in financial and healthcare algorithms. Sponsors of the bills, a bipartisan group of senators and representatives, argue that the measures are intended to prevent “rogue” AI from operating without accountability while preserving the United States’ competitive edge in technology. If passed, the bills would establish a framework for ongoing monitoring, transparency, and enforcement, marking a significant step toward federal oversight of emerging AI capabilities.