AI
Working with small teams has put me on the front line of AI - and inherently sharing the responsibility of security.
At Thoughtworks, we learnt that security is everyone’s responsibility, and security needs psychological safety and learning by everyone.
So as a non-technical person (relative to my AI/software engineering colleagues) I want to know:
What do non-technical people need to know today with the rise of Agentic AI security?
Agentic AI security
What can go wrong
From deleting the database to leaking credentials on WhatsApp groups, there are more and more admissions of what has gone wrong and could go wrong in your organisation.
The impacts are so huge - it’s important that we all know something.
We can start by knowing:
What can go wrong
What to do right
What to do right.
Limiting AI agents from access /ability of least one of the 3 in the lethal trifecta: Access to Sensitive Data, Ability to Communicate Externally, Exposure to Untrusted Content.
Korny Sietsma’s article (on Martin Fowlers website) explains this and more in a way that as a fairly technical non-technical person I can understand. (I also felt obliged to read since we share some stuff /kids).
Agent AI Security - make the right thing to do, the easiest thing to do
Stickers of the Dos and Don'ts from Korny's Agentic AI security blog - £5 for a set of 5 stickers
