Nora — AI Policy Copilot
Published by Norivo Technologies Ltd
Table of Contents
1.Model Details
type
Large Language Model Application
version
1.0
developer
Norivo Technologies Ltd
model Name
Nora — AI Policy Copilot
release Date
March 2026
underlying Model
Anthropic Claude (claude-sonnet-4-20250514)
2.Intended Use
primary
Assist compliance professionals with policy drafting, gap analysis, questionnaire responses, and impact assessment content generation
out Of Scope
Nora should NOT be used for: final compliance decisions without human review, legal advice, automated AI system risk classification without human confirmation, or any consequential decision affecting individuals' rights
3.Training Data
Nora uses Anthropic's Claude model which was trained on publicly available internet data. Nora does not fine-tune or retrain the model. At query time, Nora accesses the customer's organisation-specific compliance data (controls, evidence, policies, AI systems, risks) to provide contextual responses.
4.Evaluation
Nora's outputs are evaluated through: human review requirements on all generated content, confidence scoring on questionnaire auto-answers, user feedback mechanisms, and credit-based usage tracking to monitor adoption and quality.
5.Limitations
Nora may: generate inaccurate regulatory interpretations, miss nuances in specific jurisdictions, produce generic policy content that needs customisation, hallucinate control references or framework requirements. All outputs require human expert review. Nora does not have access to real-time regulatory updates — framework data is current as of the last platform update.
6.Ethical Considerations
Nora is classified as LIMITED risk under the EU AI Act due to direct human interaction. Transparency obligations are met through: clear AI disclosure in the chat interface, disclaimers on all generated content, this published model card, and the AI interaction badge. Nora does not process personal data beyond the user's name and organisation context.
7.Recommendations
Always review Nora's output before using it in compliance documentation. Cross-reference regulatory citations against official sources. Use Nora as a starting point for policy drafting, not as the final word. Escalate complex regulatory questions to qualified legal counsel.
8.Technical Specifications
| Metric | Value | Benchmark |
|---|---|---|
| api | Anthropic Messages API | N/A |
| rate Limits | Plan-dependent (4,000-unlimited credits/month) | N/A |
| context Window | Organisation compliance data serialised as structured context | N/A |
| response Format | Markdown text with inline control references | N/A |
EU AI Act Compliance
This model card has been prepared in accordance with EU AI Act Article 11 requirements for technical documentation. It provides transparency information about the AI system's design, development, capabilities, and limitations to support regulatory compliance and responsible AI deployment.