NLU Manifest: How Far Can Non-Generative NLP Go for LLM Governance?
Abstract
Current LLM governance systems, Llama Guard, NeMo Guardrails, and Constitutional AI, rely on large language models to monitor other large language models. This raises practical concerns around cost, latency, determinism, and auditability. We present NLU Manifest, a governance architecture that evaluates non-generative NLP as an alternative to LLM-based guards. We compare four tiers of increasing complexity: NLU-lite: SVM, TF-IDF, regex, LDA (fully deterministic,
Fields of Study & Subjects
Topic Modeling
Ethics and Social Impacts of AI
Adversarial Robustness in Machine Learning
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 4 March 2026
- Language
- EN
- Document type
- Preprint