US Lawmakers Push for Emergency 'Kill Switch' Authority Over Advanced AI Models
US lawmakers are developing legislative frameworks that would grant federal authorities the power to implement emergency shutdowns—commonly referred to as "kill switches"—on advanced artificial intelligence systems. The proposals aim to address concerns that sufficiently powerful AI models could pose national security risks if left unregulated or if they act in unexpected, potentially harmful ways.
The initiative reflects growing bipartisan concern in Washington about the pace of AI development and the potential for autonomous systems to exceed human control. Under the proposed measures, certain high-capability AI models would be subject to federal oversight, with regulators empowered to mandate safety features or, in extreme cases, order discontinuation of systems that demonstrate dangerous behaviors.
Advocates argue that such provisions are essential for managing existential risks associated with frontier AI development. Critics, however, caution that overly broad shutdown authority could stifle innovation and create uncertainty for companies investing in AI research. Industry representatives have expressed willingness to engage with safety regulations but emphasize the need for balanced approaches that address genuine risks without compromising US competitiveness in AI development.
The discussions come as AI laboratories continue to release increasingly capable models, prompting renewed calls from researchers and policymakers for concrete governance mechanisms to ensure AI systems remain aligned with human values and safety requirements.