Microsoft's CEO says he was wrong that AI would just behave itself

Microsoft CEO Satya Nadella says he used to believe AI would learn to behave as it got smarter. He no longer believes that. He shared the change of mind on a podcast this week.

What he said

On the 20VC podcast, host Harry Stebbings asked what Nadella had changed his mind about over the past year. He said he had assumed "alignment" would follow on its own as AI grew more capable. Alignment means making sure AI acts in line with human values and goals. "I used to always think that alignment would just fall out," he said. "I no longer believe that." He called it a "harder, different problem."

In an October 10 post on X, Nadella said companies should treat AI models as insider threats and assume they are "compromised." He wrote that unpredictable models need to be surrounded by strong, predictable system design, human controls and reliable operating procedures. He also called for industry standards where current ones fall short.

He is not alone. OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei have also warned about AI threats. A July letter signed by more than 1,000 AI workers said AI could advance faster than people's ability to understand or control it.

What it means

The reporting does not say what steps Microsoft is taking or which standards Nadella wants. For anyone using AI at work, the practical lesson is simple: keep a person in charge of key decisions, and limit what the tools can access.