Research
Original research on AI trustworthiness in consequential domains: hallucination patterns, evaluation methodology, citation provenance, and the architecture of accountable AI systems.
Featured
A formal knowledge representation for orally-transmitted customary law using Dempster-Shafer evidence theory, grounded in Nigerian Evidence Act doctrine.
A denotational semantics for AI agent mandates that proves scope can only narrow under delegation, as a theorem rather than a design convention.
Papers
Five structural invariants for mandate-preserving AI agent delegation chains, grounded in three centuries of common law agency doctrine.
A structured evaluation suite for LLM legal reasoning across African jurisdictions, covering Nigeria, South Africa, and Kenya.
A formal theory of how autonomous AI agents can satisfy multiple regulatory regimes simultaneously, with a three-jurisdiction case study.
A queryable, versioned, open dataset of AI regulatory obligations in machine-readable JSON, covering 45 obligations across 10 regulations.
An open JSON Schema standard for contract intelligence data, designed for portability across CLM platforms and AI extraction tools.
Detecting doctrinal drift in legal citation graphs through temporal graph networks and treatment-type-aware message functions.
A graph-aware retrieval framework for precedent search that models the citation network rather than relying on semantic similarity alone.
Every paper, benchmark, and technical report is free to read. Tensflare is also a Contributor Member of C2PA.