How to stop AI agents going rogue

Agentic AI is taking decisions and acting on behalf of users, but how to stop that going wrong?

source https://www.bbc.com/news/articles/cq87e0dwj25o?at_medium=RSS&at_campaign=rss