AI safety is designed in the West, and failing users everywhere

OpenAI’s pause shows the risks of rapid AI development, as safety frameworks fail to account for non-Western languages and contexts where harms are felt most.