Your LLM Stack is a Security Nightmare 🤖
Lessons from Hacken’s Stephen Ajayi on blockchain and AI security, LLM threats, prompt injections, and how to build safer AI systems.
Introduction
In this episode, we sit down with Stephen Ajayi, Technical Lead for dApp & AI Audits at Hacken, to talk about the dark side of large language models. From prompt injection attacks to red teaming, Stephen breaks down why your shiny new AI tools might be wide open to exploitation. If you’re building with LLMs and not thinking about security, this one’s your wake-up call.
Timestamps
01:07 Steven Ajayi's Background and Career Journey
05:09 Indirect Prompt Injection
08:31 Defending Against Prompt Injection
14:43 Building Resilient LLM Systems
17:50 Guardian AIs and Verifier AIs
20:27 AI Guardrails
23:46 Red Teaming for LLMs
29:59 Future of AI Security
38:13 Conclusion and Final Thoughts