From Refusal to Recovery: A Control-Theoretic Approach to Generative AI Guardrails

Publication
(in submission) arXiv preprint 2025