Microsoft CEO Nadella No Longer Thinks AI Alignment Will Fix Itself

Microsoft CEO Satya Nadella has changed his view on AI "alignment". On the 20VC podcast, he said he used to believe that the more powerful AI systems get, the easier it is to keep them in line with human values. He no longer thinks the two will naturally move forward together.

Nadella explained that as models grew bigger, AI kept getting better, but the alignment problem stayed unsolved. He said it may be a tougher issue of a different kind. In a post on X, he urged companies to treat AI models as internal security threats, and build safeguards with reliable system design, human oversight and industry standards.

The AI safety topic is getting wide response across the tech industry. OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei have both warned about AI risks. In July, more than 1000 AI workers signed an open letter, saying AI may grow faster than humans can understand or control.