Saffron Birch is a fourth‑year Computer Science student specializing in ethical AI, model governance, and human‑AI interaction. Her research focuses on developing mindful AI frameworks that ensure models operate within clearly defined behavioural boundaries, with an emphasis on guardrails and responsible deployment.

She works at the intersection of machine learning, HCI, and AI ethics, designing systems that are transparent and aligned with human values.

She recently completed her undergraduate thesis, in which she developed an adversarial framework to evaluate the performance of LLM-driven non-player characters (NPCs) in video games. The framework uses an adversarial test suite designed to provoke role-breaking behaviour. It monitors an NPC’s candidate response to an adversarial question to determine whether it successfully reflects the character’s outlined persona and maintains consistent behaviour grounded in their source material.