1 paper · 1 filter
Emily Broadhurst, Tawab Safi, Joseph Edell +2
Conversational AI systems require guardrails to prevent harmful outputs, yet existing approaches use static rules that cannot adapt to new threats or deployment contexts. We introd…