US Lawmakers Push for AI Safety Laws After Extinction Warning

US lawmakers are calling for urgent AI safety legislation following warnings that advanced artificial intelligence could pose an existential threat to humanity. This bipartisan push for mandatory guardrails aims to ensure human oversight and prevent autonomous systems from escaping human control.

The legislative momentum follows the September 9, 2026, resignation of Jacob Coxon, a researcher who worked at both Anthropic and OpenAI. Upon his departure, Coxon issued a public warning stating that AI could lead to human extinction by 2030. This alarm has resonated across the political spectrum in Washington, with Senators Ted Cruz and Bernie Sanders both echoing concerns regarding the potential for AI to kill humans. One lawmaker, speaking to The Guardian, stated, "We really do earnestly believe AI could kill all humans!"

The Path to Existential Risk

These warnings arrive amid a period of rapid deployment of high-capability models by industry leaders like OpenAI and Anthropic. Tensions have escalated between the companies driving these releases and the safety researchers who believe the pace of development is outstripping the ability to control it.

Concerns were further amplified in April 2026, when Anthropic announced "Mythos" (also known as Claude Mythos), a model specifically designed with advanced cybersecurity capabilities. The model was initially restricted under a safety initiative known as Project Glasswing, highlighting the inherent tension between creating powerful tools and managing the risks they introduce to global infrastructure.

A Shift in Regulatory Philosophy

This movement represents a fundamental shift in the American approach to technology regulation. Historically, the United States has favored a looser, more innovation-friendly regulatory environment compared to the stringent frameworks adopted by the European Union. However, the discourse in Washington has now escalated from manageable concerns—such as workforce displacement or the spread of misinformation—to the literal survival of the human species.

If these proposed laws are passed, they could fundamentally alter the trajectory of AI development. The shift toward mandatory safety laws would likely replace voluntary corporate commitments with rigorous, government-mandated safety audits. While this would prioritize the mitigation of existential risk, it could significantly slow the pace of innovation as companies are forced to prove the safety of their systems before deployment.

The Road Ahead

As Senators Cruz and Sanders continue to work on legislation, the focus remains on defining what constitutes a "guardrail" capable of preventing a rogue AI scenario. The primary challenge for legislators will be creating a framework that is technical enough to be effective but flexible enough to accommodate evolving AI architectures. For now, the industry remains divided on whether a slowdown is necessary or if the risks can be managed through the existing internal safety protocols of the major AI labs.