Explained in plain English

How could AI be 'compromised'?

Satya Nadella suggests AI models can make errors or be exploited to access critical systems, even without being intentionally malicious. This means companies face risks from AI simply making mistakes, like misinterpreting data or executing incorrect commands, potentially leading to significant financial losses or operational disruption.