California pushes for AI safety measures including potential kill switch

California Governor Gavin Newsom has issued an executive order to establish an expert panel focused on enhancing AI safety laws, including the possibility of an 'emergency shut off' for advanced AI systems, known as an 'AI kill switch.' The order aims to accelerate third-party safety evaluations and mandatory risk reporting. Travis Hall of the Center for Democracy and Technology emphasized the need for diverse expertise in these processes, including civil rights advocates. The concept of a kill switch has gained traction in legislative discussions, with the bipartisan AI Kill Switch Act introduced in July, granting the Department of Homeland Security authority to halt AI systems deemed capable of causing catastrophic harm. While most AI labs have internal safety protocols, experts argue that binary safety regulations may be ineffective and prone to abuse. Hall warned that such measures could introduce new risks and urged for ongoing, independent oversight. Experts like Mark Nitzberg and Geoffrey Hinton have questioned the feasibility of a universal kill switch, noting that AI systems operate in complex environments that cannot be easily contained. Michael Vermeer suggested that isolating rogue AI may require a global effort. Ultimately, the technology's control remains in human hands, raising questions about who should have the authority to act in case of a crisis.