AI agents asking questions in natural language apparently issue a lot more queries than your average SQL jockey ...

For CISOs evaluating AI, the key question is whether a model truly understands the environment it operates in.