@ericyocamProfessor, California Polytechnic State University
How Are You Securing Your Agentic AI Systems?
Which tool or approach are you using to secure your Agentic AI systems against prompt injection and unauthorized actions?
@ericyocam @ericyocam
1 comment
For me it's mostly least-privilege: agents only get the tools they need, plus my approval on anything that sends, deletes or spends money. I'd vote for that option.