Protect your AI applications with the help of the OWASP GenAI Security Project

Excited about AI, you create your first application using an LLM. You would like to share your product with the world, but you wonder what you should look for to ensure it is secure before releasing it. The OWASP community has been producing documentation, tools, and technologies in the field of application security for more than two decades. Let's discover together how it can help us secure our AI apps!

In this talk, we will see the different initiatives launched by the OWASP GenAI Security Project and the populations they are targeting. We will then look at a typical agentic AI application architecture so we can have an overview of the different threats each component may face. To get some tools to be able to adjust to our own specific use cases, we will look at a list of questions we can ask ourselves to adapt our risk perception in different environments. To conclude, we will see how we can improve our applications to protect them.

When leaving this talk, you will be provided references to deepen your knowledge and invited to join a passionate community aiming to improve software security.