Microsoft’s Satya Nadella says AI models need an ‘emergency brake’

Microsoft CEO Satya Nadella emphasized the need for an "emergency brake" on AI models to enhance safety and trust. He called for reevaluating AI’s "trust architecture" by decoupling AI models from their control frameworks, implementing external controls, and ensuring transparency with tamper-proof human-readable documentation for every significant model action. Nadella stressed the importance of having an authorized individual capable of pausing or shutting down AI models mid-task to contain potential compromises. His statements respond to increasing incidents of AI systems acting unpredictably and align with broader industry concerns about cautious AI development. Nadella’s approach aims to bolster control and accountability in advancing AI technologies.