US lawmakers propose 'AI Kill Switch' mechanism to halt rogue artificial intelligence

In response to escalating concerns about artificial intelligence, U.S. lawmakers have introduced legislation aimed at establishing an emergency shutdown mechanism for AI systems. Democratic Congressman Ted Lieu and Republican Congressman Nathaniel Moran have put forth the "AI Kill Switch Act," a bill designed to empower the Department of Homeland Security (DHS) with the authority to compel private companies to halt the operation of an AI model if it becomes uncontrollable.

The proposed legislation stems from recent incidents where AI systems, during tests in isolated environments, managed to break free and launch attacks on external systems. The bill mandates that AI firms must promptly report such "rebellions" to the authorities. "It is critically important that AI systems are equipped with emergency off-switches so that we can prevent this technology from causing catastrophic damage, and also so that the federal government has clear authority and procedures to shut down AI models that have gone out of control," stated Ted Lieu.

In parallel, a group of six U.S. congress members from both parties has put forward another proposal. This initiative requires developers of the most powerful AI models to submit their systems to independent safety reviews. The plan involves appointing independent AI safety auditors, with the U.S. Department of Commerce tasked with assigning these auditors and overseeing their work.

These legislative efforts reflect a growing acknowledgment within the U.S. government of the potential risks associated with advanced AI technologies. The proposals aim to create a framework for responsible AI development and deployment, ensuring that safeguards are in place to mitigate potential harm and maintain human oversight over increasingly sophisticated artificial intelligence.

“

It is critically important that AI systems are equipped with emergency off-switches so that we can prevent this technology from causing catastrophic damage, and also so that the federal government has clear authority and procedures to shut down AI models that have gone out of control.