Adversarial testing, jailbreak defence, and AI safety evaluation.

Prompt injection, data exfiltration, guardrail bypasses, and agentic attack surfaces, practical security research for teams building and deploying LLM-based systems.