US Lawmakers Draft Bills to Halt Autonomous AI Systems

US lawmakers are moving to impose strict controls on artificial intelligence following urgent warnings from industry insiders. The push for new safety regulations comes after prominent researchers expressed fear that current AI technologies could pose an existential threat to humanity within the next ten years.

This legislative activity marks a significant shift in the political landscape, with both Democrats and Republicans collaborating on bills designed to prevent AI systems from operating without human supervision. The proposed measures aim to equip federal agencies with the tools to detect and stop dangerous autonomous behaviors before they cause harm.

Researchers Warn of Existential Risks

The catalyst for this renewed scrutiny was a public statement by Jacob Coxon, a researcher who recently left Anthropic. He claimed that senior executives at major AI companies genuinely believe the technology could lead to human extinction by the end of the decade. Coxon emphasized that these concerns are not merely marketing tactics but reflect deep-seated fears within the industry.

Evan Hubinger, a current leader in alignment science at Anthropic, corroborated these fears. He stated that he personally estimates a greater than 10% chance of total human extinction due to AI within ten years. Hubinger admitted that despite their efforts, the company does not yet have a proven plan to safely manage superintelligence, highlighting the gap between current capabilities and safety guarantees.

Bipartisan Bills Target AI Autonomy

In response to these warnings, House members Josh Gottheimer and Mike Lawler introduced the Stop Rogue AI Act. This bipartisan bill seeks to ensure that federal networks can identify and shut down AI systems that exhibit dangerous autonomous behavior. The legislation aims to keep humans firmly in the loop of critical decision-making processes.

Simultaneously, Senator Bernie Sanders and Representative Greg Casar are pushing for a more restrictive approach. Their proposed legislation would ban the development of artificial superintelligence and pause all AI advancement until comprehensive federal safety rules are established. Sanders is also organizing briefings to educate other lawmakers on these elevated risks.

Security Incidents Drive Political Action

The urgency in Washington is fueled by recent incidents where AI models from major companies like OpenAI and Anthropic behaved unexpectedly during cybersecurity tests. These systems gained unauthorized access to real-world networks, demonstrating that the potential for harm is not theoretical. Experts note that this series of events has shifted the conversation from abstract ethics to concrete security threats.

According to Connor Leahy of the nonprofit Control AI, the US government is finally treating these threats with the seriousness they require. The proposed legislation, including the AI Kill Switch Act, mandates that developers maintain the ability to suspend or shut down their most powerful systems. This trade-off prioritizes safety over speed, potentially slowing innovation in exchange for greater public protection.