From below:

"What these researchers are creating [AI Agents] could soon escape our control.
And that would be really bad."

My comment:
They already have (escaped our control).
Study the Hugging Face Incident where you'll find the answers.