ai
Implementing Computational Law in Wolfram Language for the Governance of Artificial Intelligence
- Source
- arXiv — Computers and Society
- Published
- Last verified
- 18 Aug 2026
- Confidence
- High
- Evidence
- Original document retained
- Reading time
- 1 min
- Country
- International
- Relevant to
- Research & Evidence, Operations & Delivery, Technology & Data, Board & Governance, Risk & Compliance
Executive summary
What happened, and why should leadership care?
This research explores a method for governing Artificial Intelligence (AI) systems through "Computational Law," specifically implementing Reified Input/Output Logic in Wolfram Language. The approach focuses on defining and verifying AI compliance with explicit obligations, permissions, and prohibitions, rather than attempting to inspect complex internal reasoning. Initial tests using GPT-4 to translate English legal statements into this formalism revealed significant challenges, including the generation of incorrect or subtly flawed code, highlighting the complexities in automating legal interpretation for AI governance.
Why this matters
Why is this strategically important?
The increasing complexity and opacity of AI systems necessitate novel governance mechanisms that can ensure compliance without requiring complete transparency into internal processes. This research offers a structured approach to defining and enforcing AI behavior through formal computational law, which is crucial for building trust and managing risk. Addressing the identified challenges in translating legal text to computational forms will be vital for the practical deployment of such governance frameworks.
Key insights
What should be noted from the evidence?
- AI governance does not necessarily require full inspection of an AI system's reasoning, but rather clear definitions of its obligations, permissions, and forbidden actions.
- The Reified Input/Output Logic, fundamental to the DAPRECO knowledge base, can be implemented within the Wolfram Language.
- Key components implemented include core I/O axioms, obligations, permissions, constitutive norms, reified eventualities, and temporal operators.
- Initial testing with GPT-4 to translate English legal statements into the computational formalism demonstrated failures, including hallucinated functions, omitted temporal scope, deviations from the formalism, and plausibly-reading but silently incorrect code.
- The development of robust computational law for AI governance faces significant hurdles in accurately translating natural language legal concepts into formal logic.
Evidence and confidence
How far can this assessment be trusted?
High confidence. Named institution, original document retained and analysis corroborated.
Analysis is prepared editorially by Aziz Shuaib Ausi. The original publication remains the authoritative record, and executive judgement remains entirely human.
Source
Where does this originate?
Reported by arXiv — Computers and Society · International. This briefing summarises the publication for executive use; the document itself is not reproduced here.
Read the original publication