New Delhi: Microsoft CEO Satya Nadella has cautioned organisations against placing unquestioning trust in artificial intelligence (AI), as advanced models become more widely used across sectors.
He said companies should stop treating AI outputs as automatically reliable. When deploying powerful, next-generation AI systems, firms must build control mechanisms—akin to an “emergency brake”—to intervene if something goes wrong.
Nadella stressed that qualified humans should have the authority to pause or disable an AI system mid-task when errors or risks emerge. He also warned organisations not to rely solely on assurances made by companies that develop AI models.
Even when using highly sophisticated AI, he said it should be approached as potentially unsafe or threatening by default and kept within strict guardrails from the outset. He compared the risk to an insider employee who could steal or misuse confidential company information, and added that AI should not become the final authority replacing human decision-making.
He further noted that regardless of how advanced a “super intelligence” may be, its operations must be documented in a way that humans can read and understand. With AI increasingly dominating multiple fields, his remarks have drawn global attention.
