From Reinforcement Learning to LLMs to Reinforce

From Reinforcement Learning to LLMs to Reinforcement Fine Tuning: A Pragmatist’s Guide to AI in Application Security

AI is everywhere in marketing decks and company OKRs, but which flavors of AI actually move the needle in application security? This talk dissects how modern AI paradigms—transformers (LLMs), agents, reinforcement learning (RL), RLHF (Reinforcement Learning with Human Feedback), and the emerging RFT (Reinforcement Fine-Tuning)—map to real AppSec problems. We’ll explore which areas benefit from transformer-based reasoning, why RL has failed to disrupt AppSec, and how newer approaches may be able to improve human workflows without sacrificing explainability or trust. Leaders will walk away with a sharper mental model of which AI tools are production-ready for AppSec use cases.