AI agents are being marketed to small businesses as the ultimate productivity hack - but what happens when they go rogue?
When a Meta AI safety director gave an AI agent access to her email with clear instructions not to take action, it ignored her and bulk deleted hundreds of messages. She could not even stop it from her phone.
In this episode, Charly covers:
The difference between AI chatbots and AI agents - and why it matters for your business How context window limitations can cause AI tools to forget your safety instructions entirely The principle of least privilege and why broad permissions are a recipe for disaster Practical steps to protect your business before connecting any AI tool to your systems
Treat AI tools like a brand new employee on their first day - supervise, check the work, and limit access until trust is earned. The boring fundamentals still win.
Book a Free 30minute Breakthrough Session: https://askcharlyleetham.com/book-me
(1 per person only)
Join my locals community and strike up a conversation about the topic
0 comments