An AI question for the experts.

An AI question for the experts.

With all the recent buzz about AI Agents, does this not imply I need to hand over an amount of trust to the agent to perform anything of value?

E.g "please look at my calendar , book flights and accomodation for the next 3 business trips, trips should be no more than $X.

What's the thinking / approach to trust boundaries and limitation of risk in an AI agent world?