An open framework by VisionZeroAI (Pty) Ltd
Protecting private and confidential data from unauthorized AI ingestion.
The AI-Restricted Open Standard (AIROS) defines a universal tagging and verification framework that allows organizations to mark digital documents as “AI-Restricted.”
This standard ensures that sensitive or private files — such as internal reports, client data, or IP-sensitive materials — cannot be uploaded or ingested into AI systems, LLMs, or automated data pipelines without consent.
AIROS enables cross-platform interoperability so that AI vendors, enterprises, and cloud providers can detect and respect AI-restriction flags embedded within document metadata or sidecar files.
At VisionZeroAI, we believe privacy and innovation can coexist.
AIROS was developed to:
- Prevent unauthorized AI ingestion of sensitive data.
- Empower organizations to control how and where their documents are used.
- Encourage responsible AI adoption through verifiable, transparent standards.
- Build a trusted global ecosystem where AI systems respect digital boundaries.
AIROS introduces a verifiable, vendor-neutral tagging system for digital assets.
Core Components
- Metadata Tagging (XMP / Custom Properties) — embeds cryptographically verifiable AI-restriction identifiers within supported file formats (PDF, DOCX, TXT, etc.).
- Verification SDKs (Python, Node, REST) — allow AI platforms to auto-check for AIROS tags before accepting or processing uploads.
- Trust Registry (Planned) — a global directory for key verification, revocation, and organization identity validation.
Future Integrations
- Cloud DLP and compliance systems (e.g., Microsoft 365, Google Workspace, AWS Macie).
- AI platform ingestion filters (OpenAI, Anthropic, Hugging Face, etc.).
- Document lifecycle monitoring and compliance analytics.
| Sector | Use Case |
|---|---|
| Enterprises | Protect intellectual property and client data from unauthorized AI uploads. |
| Governments | Enforce data residency and privacy mandates automatically. |
| AI Companies | Demonstrate ethical data handling through AIROS compliance. |
| Developers | Integrate AIROS verification into upload pipelines or document processors. |
| Document | Description |
|---|---|
| AIROS Technical Framework v1.2 (Whitepaper PDF) | Full technical specification of the AI-Restricted Open Standard. |
- Version: 1.2 (Technical Framework)
- Maintainer: VisionZeroAI (Pty) Ltd — South Africa
- Author: Frans Werner Labuschagne
- License: Creative Commons Attribution–NonCommercial 4.0 International (CC BY-NC 4.0)
AIROS welcomes collaboration from:
- AI platform providers
- Security and compliance solution vendors
- Academia and standards organizations
- Privacy and governance institutions
Interested in contributing or partnering?
📧 info@visionzero.co.za
You can also participate through our GitHub Issues or by submitting Pull Requests.
This project and all related documentation are licensed under the
Creative Commons Attribution–NonCommercial 4.0 International License (CC BY-NC 4.0)
© 2025 VisionZeroAI (Pty) Ltd — All rights reserved.
Author: Frans Werner Labuschagne
Website: https://www.visionzero.co.za
ai-security • data-protection • privacy • open-standard • ai-compliance • visionzeroai • airos
Help shape the future of responsible AI data governance —
Adopt the AI-Restricted Open Standard (AIROS).