# AI provisions reviewer
> Source: https://legalcode.md/skills/legalcode-ai-provisions-reviewer | [HTML version](https://legalcode.md/skills/legalcode-ai-provisions-reviewer)

Reviews contracts for AI-specific provisions: IP ownership of AI-generated outputs, AI usage restrictions and acceptable-use policies, training data rights…

Reviews contracts for AI-specific provisions: IP ownership of AI-generated outputs, AI usage restrictions and acceptable-use policies, training data rights (inputs/outputs as training material), liability allocation for AI errors and hallucinations, transparency and explainability requirements, algorithmic audit rights, bias monitoring and disparate impact obligations, human oversight and human-in-the-loop requirements, AI governance and responsible-AI clauses, and missing AI governance terms. Flags contracts that lack adequate AI safeguards, identifies one-sided AI risk transfers, generates redlines, and produces a confidence-scored, auditable AI governance analysis. Use when reviewing SaaS agreements, professional services contracts, technology licenses, vendor contracts, employment agreements, or any commercial contract where either party uses, deploys, procures, or integrates AI systems. Applies EU AI Act 2024/1689, GDPR Art. 22, NIST AI RMF, NYC Local Law 144 (AEDT), Colorado SB 205 (2024), US Copyright Office AI guidance (Jan 2025), and UK IP Office AI position. Jurisdiction-agnostic with [JURISDICTION-SPECIFIC] markers. Severity classification: GREEN / YELLOW / RED / [AI-GOVERNANCE-GAP].

This skill reviews existing contracts for AI-related provisions, identifies gaps in AI governance coverage, classifies deviations by severity, generates actionable redlines, and produces a confidence-scored, auditable analysis. **Covers:** - IP ownership of AI-generated outputs and AI-assisted work product - AI usage restrictions, acceptable-use policies, and tool prohibitions - Training data rights: using inputs, outputs, or data to train, fine-tune, or improve AI models - Liability allocation for AI errors, hallucinations, and autonomous AI decisions - Transparency, explainability, and algorithmic audit rights - Bias monitoring, disparate impact obligations, and algorithmic accountability - Human oversight requirements: human-in-the-loop, meaningful human control - AI governance: responsible AI clauses, incident reporting, model cards, ethics by reference - AI data protection overlays: GDPR Art. 22 automated decision-making, DPIA triggers - Vendor AI supply chain: sub-processor AI use, model provider dependencies, deprecation risk - Missing AI governance terms: flagging contracts that are silent on AI entirely **Does not:** - Draft new AI contracts (use a drafting-specific skill) - Provide legal advice or replace qualified counsel - Apply exclusively to one jurisdiction — jurisdiction-agnostic with [JURISDICTION-SPECIFIC] markers - Assess the underlying AI system for technical safety (a separate engineering and risk function) This is a jurisdiction-agnostic skill. The ...

## Jurisdictions

Global



---
*Legalcode -- legalcode.md*