In this episode of the First Impressions Podcast, FIRSTCON26 Speakers, Cheng-Lin Yang and Lily Chen of CyCraft, explore a growing threats to LLM-assisted security workflows: poisoned artifacts designed to manipulate AI-generated analysis. They discuss how a single malicious document can attract an AI's attention, influence its reasoning, and ultimately alter the information presented to security teams. The conversation covers the evolution of prompt injection attacks, the challenges of defending modern language models, and more! Tune in to preview their FIRSTCON26 session.

Rss Apple Podcaster
Podden och tillhörande omslagsbild på den här sidan tillhör FIRST.Org. Innehållet i podden är skapat av FIRST.Org och inte av, eller tillsammans med, Poddtoppen.