PRIME Guardrails: A General, Low‑Latency Safety Framework for Generative AI

Afshaan Mazagonwalla, Yiran Si

Published: 22 Sept 2025, Last Modified: 03 Jan 2026WiML @ NeurIPS 2025EveryoneRevisionsBibTeXCC BY 4.0
Keywords: GenAI Safety, Safety guardrails, Prompt Injection, Text Safety, Image Safety, Multimodal Safety, Bias, Misinformation, Safety Framework
Submission Number: 354