Abstract
A system can address and provide a hybrid evaluation model that applies a principle of monotonic fusion for automated enterprise AI risk governance and compliance gateway control across corporate risk tiers. The system may perform a deterministic assessment to calculate a baseline risk score, or risk floor, and in parallel, may use a large language model for a probabilistic semantic interpretation of unstructured descriptions. A final risk classification can be determined by taking the maximum of the deterministic floor and the model-generated score, formally modeled as a monotonic lattice join over a bounded risk lattice. This architectural design may be configured such that the probabilistic component does not de-escalate the risk classification below the deterministic baseline, which can help mitigate incorrect risk reduction, adversarial prompt injection bypasses, or omission hallucinations. Continuous runtime compliance is enforced via cryptographic attestation digests validated by a co-located execution daemon.
Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Enoch, John, "Monotonic Fusion of Deterministic and Probabilistic Assessments for AI Risk Governance", Technical Disclosure Commons, (November 20, 2026)
https://www.tdcommons.org/dpubs_series/13187