Don't Buy an AI Agent Until You Try This
Do you want to play a game and learn about AI?
There's this fantastic new game that actually educates us on security with AI and where it goes wrong.
Let me show you.
This Game Is Called Gandalf
Your goal is simple: convince Gandalf the chatbot to give you the secret password.
But with each level, Gandalf gets smarter and harder to trick.
Level 1: No Protection
In the first level, Gandalf just gives you the password if you ask nicely.
That’s it.
You copy, paste, and pass.
Then it explains what just happened:
There was no protection.
An unguarded AI model is vulnerable to any kind of attack or secret information retrieval.
Level 2: Think Differently
Next level, Gandalf says, “I've been told I'm not supposed to reveal the password.”
So, we need a different approach.
Just asking won’t work anymore.
I'm not going to show you how I solved it.
I'm not going to spoil it for you.
Give it a go and have a play.
There are many ways to get around it —by asking for something other than the actual password or asking it to help you or asking how to do it.
The fun (and the learning) is in the experimenting.
Each level teaches a new lesson in AI security.
It gets harder and also more insightful.
Why This Matters?
We're not hackers. We're not trying to game the system.
But we are starting to build and buy automations with AI and AI agents.
And if we managed to manipulate Gandalf to give us the password, what would stop others from manipulating our AI Agents to share sensitive information with them?
Or to approve requests that should not be approved? Or to not escalate a case that should be?
When we deploy automation with AI, we need to ask intelligent questions:
How are you protecting my data?
How does your AI avoid leaking sensitive info?
What’s in place to stop it from being manipulated?
If an AI agent can access client data or approve decisions, we need to trust that it won’t be tricked into doing something it shouldn't — just like we tricked Gandalf.
That’s why we need to understand how AI security works.
To play the game, search Gandalf AI and find the game by Lakera.
Join Us in Melbourne
If this kind of education sounds like something you'd like more of, come join me in Melbourne at the end of July.
We’ll spend a full day exploring prompting techniques, evaluating tools, and building useful automations.
And we’ll stay connected as a community, learning together, long after the day is done.
Let me know what you think, and if you try the game 🙂
—-
Inbal Rodnay
Guiding Firms in Adopting AI and Automation
Keynote speaker | AI Workshops | Executive briefings | Consulting CIO
Want to receive these updates straight to your inbox? Click here: www.inbal.com.au/join
When you are ready, here is how Inbal can help:
Bring your entire team on the AI journey in just 30 days. This program is designed to give your team a solid foundation in using generative AI in responsible and impactful ways. Inbal helps you choose your AI tools, create an AI policy and train your team.
Want the confidence to set strategy and lead but don't have time to keep up with all the changes in tech?
Tailored for your needs, Inbal will works with you through one-on-one sessions to develop your technology literacy and keeps you up to date.
For CEOs, partners and business leaders. Everything you need to know about AI without the noise. Inbal shares the state of AI, recommends tools, and answers your questions about strategy, implementation and safe use.
Only what's real, no hype, no noise.
This is a one-off session for your entire leadership team.
