Microsoft CEO Reveals AI's Biggest Deception: Users Pay Twice

Jakarta, CNBC Indonesia - As artificial intelligence has developed rapidly in recent years, many have grown concerned about its risks. Microsoft CEO Satya Nadella has now joined those voicing apprehension.

According to him, AI users are paying twice: once through usage tokens and, unknowingly, again with their own data.

“You are essentially paying twice, once with money and once with something more valuable: your proprietary knowledge that makes the intelligence useful,” Nadella said in a blog post, as quoted by Tech Crunch on Tuesday (14/7/2026).

He added that this “payment” increases when users want the model to perform better.

“The better you want the model to perform, the more knowledge you have to give it,” he added.

He further wrote that companies teach AI models about their businesses in great detail. Every prompt and correction made while using AI becomes new knowledge.

Nadella also argued that AI activities must be conducted in a balanced manner. Currently, model owners can train their systems on data scattered across the world while restricting others from doing the same with their models.

Additionally, he expressed concern that AI model creators have the right to learn from customer usage data and interactions.

Regarding solutions, Nadella stressed that companies must retain ownership of their data, including prompts and feedback. Businesses should build exclusive learning environments in the cloud and orchestration layers, enabling them to switch between AI models from different providers.