Legal Information
Licensing terms for the vfairness library and the validant.ai AI Fairness Assessment and Interventions Platform.
On This Page
vfairness Library Apache-2.0
Copyright © 2026 Glinz & Company GmbH
vfairness is free and open-source software, licensed under the
Apache License, Version 2.0. You may use, reproduce, modify, and
distribute it, including in commercial and closed-source products, subject to the
terms of the license. Apache-2.0 adds an explicit patent grant and requires that
the LICENSE and NOTICE files be retained in redistributions.
The full license text ships in the LICENSE file with every
distribution, and attribution is recorded in NOTICE. The library is
provided as-is, as free and open-source software: there is no warranty and no
service-level or support commitment attached to it.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
AI Fairness Assessment and Interventions Platform
© 2026 iceberg.digital / Glinz & Company GmbH. All rights reserved.
The AI Fairness Assessment and Interventions Platform has been developed by the digital trust research group iceberg.digital, hosted by Glinz & Company GmbH.
The methodological foundations originate from the DIVERSIFAIR Project, an EU-supported initiative promoting fair, transparent, and accountable AI.
Openly Licensed Methodology CC BY-NC-SA
The methodology incorporated into the AI Fairness Assessment and Interventions Platform is accessible via iceberg.digital and licensed under the Creative Commons BY-NC-SA (Attribution–NonCommercial–ShareAlike) license.
License Elements
What You Can Do
Two conditions apply to everything above: NonCommercial (the material may not be used for commercial purposes without a separate agreement) and ShareAlike (adaptations must be distributed under the same CC BY-NC-SA license). Note that CC BY-NC-SA is an open-content license with a NonCommercial restriction; it does not meet the Open Source Definition. The vfairness library itself is separate and is genuinely open source under Apache-2.0, as described above.
Commercial Platform Proprietary License
While the underlying methodology is open-source, the AI Fairness Assessment and Interventions Platform, as a complete software product, includes proprietary tools, software modules, integrations, and implementation findings developed by iceberg.digital.
Proprietary Components
The following components are not covered by the CC license:
- Custom software tools and modules
- Platform integrations and APIs
- Implementation findings and best practices
Commercial Licensing
The full commercial software platform is available for commercial use under a paid subscription (Validant Platform Pro, billed monthly or yearly). Current plans, and the sponsorship tiers that include Platform Pro, are described on Our Offering. For other licensing arrangements, contact us below.
Sponsorship & Pro Terms
These terms govern the vfairness sponsorships and the vfairness Pro supporter subscription described on Our Offering. Sponsorship and Pro checkout are not open yet; they launch with the beta. The terms below describe what will apply when they open.
Patronage, not a purchase
A sponsorship or Pro subscription is support for an open-source project, not the purchase of software, features, or guarantees. The complete vfairness library is free for everyone under Apache-2.0 regardless of sponsorship, and sponsoring does not unlock any functionality.
No SLA, no warranty
No sponsorship tier carries a service-level agreement, a guaranteed response time, a warranty, or any legal or contractual assurance about the software. The library remains provided as-is under the Apache-2.0 warranty disclaimer above. Support stays community-based (GitHub Issues) at every tier.
Cancellation
Sponsorships are recurring and can be cancelled at any time.
Recognition
Sponsors can be named on the offering page and in the project credits, and Corporate Patrons can feature a company logo and receive a named backer credit. Recognition is our thank-you, extended in good faith; it is goodwill, not a contractual deliverable, and anonymity is always respected on request.
Platform Pro inclusion
A sponsorship of USD 50 per month or more (USD 25 during the beta discount) includes a Validant Platform Pro membership at no extra charge: after linking the sponsor account on validant.ai, Platform Pro unlocks automatically and lapses when the sponsorship stops. The platform is a separate, proprietary product; its own terms apply to its use.
Legal Disclaimer
Not Legal Advice
The vfairness library and the AI Fairness Assessment and Interventions Platform are technical tools for measuring statistical properties of machine learning models. The metrics, thresholds, and recommendations provided by this software do not constitute legal advice regarding compliance with anti-discrimination laws, regulations, or standards.
User Responsibility
Users are solely responsible for:
- Determining the appropriateness of fairness metrics for their specific use case
- Setting thresholds that meet their legal and regulatory requirements
- Interpreting results in the context of their domain and jurisdiction
- Consulting with qualified legal counsel regarding compliance obligations
No Warranty of Fairness
Passing fairness metrics computed by vfairness or the platform does not guarantee that a system is "fair" in any legal, ethical, or common-sense meaning of the term. Fairness is a complex, context-dependent concept that cannot be fully captured by any set of statistical measures.
Regulatory Compliance
References to regulations (EU AI Act, EEOC guidelines, etc.) in this documentation are for informational purposes only. Regulatory requirements vary by jurisdiction and are subject to change. Users should verify current requirements with appropriate authorities.
AI-Assisted Development Disclosure
AI-Assisted Development
In the spirit of transparency, we disclose that this project has been developed with the assistance of AI-powered coding tools. The development team has utilized the following AI technologies:
AI pair programming assistant for code completion and suggestions
Primary AI assistant for code generation, documentation, and architecture (Claude Opus was used for earlier development)
Agentic coding assistant for complex implementation tasks
AI system for code generation and natural language to code translation
Human Oversight Statement
AI tools draft; humans direct and answer for the result. The development team maintains responsibility for:
- Architectural decisions and system design
- Code review and quality assurance
- Security auditing and vulnerability assessment
- Testing, validation, and verification
- Final approval of all committed code
To be precise about the scope of that oversight: given the breadth of the library, we do not claim a line-by-line human review of every generated line. What we do claim is that every release must pass the automated quality gates described on the Quality & Hardening page, including the test suite, lint and type checks, security scanning, and cross-library metric parity checks, and that a human approves what ships.
Why We Disclose This
As a project focused on AI fairness and transparency, we believe in practicing the transparency we advocate. AI-assisted development tools are becoming standard in software engineering, and we choose to openly acknowledge their role in our development process. This disclosure reflects our commitment to honest communication about how modern software is built.
Third-Party Licenses
vfairness depends on the following open-source libraries. The core dependencies install with the package; everything else is an optional extra (installed as, for example, pip install "vfairness[xai]") and is only pulled in when you ask for it. The list is generated from the dependency groups declared in pyproject.toml, and each license was verified against the package's own distribution metadata, or the PyPI registry where the package was not installed locally (last regenerated 2026-08-22). All of these licenses permit commercial use. The explainability stack in particular was chosen to stay MIT/BSD: Alibi was deliberately excluded after its relicense to the Business Source License.
| Library | License | Component |
|---|---|---|
| NumPy | BSD 3-Clause (vendored parts under 0BSD, MIT, Zlib, CC0-1.0) | core |
| pandas | BSD 3-Clause | core |
| SciPy | BSD 3-Clause | core |
| requests | Apache-2.0 | core |
| scikit-learn | BSD 3-Clause | core |
| Matplotlib | Matplotlib license (PSF-based) | viz extra |
| seaborn | BSD 3-Clause | viz extra |
| Jinja2 | BSD 3-Clause | rendering extra |
| Plotly | MIT | dashboard / interactive extras |
| Dash | MIT | interactive extra |
| PyWavelets | MIT and BSD 3-Clause | monitoring extra |
| PyTorch | BSD 3-Clause (bundled parts under Apache-2.0, BSD 2-Clause, BSL-1.0, MIT) | training / llm-transformers / xai extras |
| MLflow | Apache-2.0 | mlops extra |
| Weights & Biases (wandb) | MIT | mlops extra |
| sentence-transformers | Apache-2.0 | embeddings extra |
| vaderSentiment | MIT | llm extra |
| alt-profanity-check | MIT | llm extra |
| Transformers (Hugging Face) | Apache-2.0 | llm-transformers extra |
| Flair | MIT | llm-transformers extra |
| Detoxify | Apache-2.0 | llm-transformers extra |
| SHAP | MIT | xai extra |
| LIME | BSD | xai extra |
| DiCE (dice-ml) | MIT | xai extra |
| Captum | BSD 3-Clause | xai extra |
| anchor-exp | BSD | xai extra |
| sage-importance | MIT | xai extra |
| DoWhy | MIT | causal extra |
| NetworkX | BSD 3-Clause | causal extra |
| mcp (Model Context Protocol SDK) | MIT | mcp extra |
| cryptography | Apache-2.0 or BSD 3-Clause (dual) | pulse extra |
| certifi | MPL-2.0 | pulse extra |
| supabase | MIT | worker extra |
| pytest, pytest-cov, Ruff, mypy | MIT | dev extra (contributor tooling) |
| Hypothesis | MPL-2.0 | dev extra (contributor tooling) |
Privacy & tracking
This documentation site uses only functional local storage (to remember your theme choice). It sets no tracking cookies and loads no third-party analytics. External requests to two third parties do occur when a page loads: the Inter and JetBrains Mono typefaces are fetched from Google Fonts (fonts.googleapis.com and fonts.gstatic.com), which transmits your IP address and user agent to Google, and a consent script is loaded from validant.ai. Beyond that, browsing these docs sends no personal data to us. The vfairness library itself runs offline by default and contains no telemetry.
Full company and legal details (registered address, commercial register, UID) are in the validant.ai imprint. The privacy notice, which also covers the platform and the vfairness library, is at validant.ai/privacy.
Contact Information
For licensing inquiries, commercial partnerships, or legal questions:
Glinz & Company GmbH
Operating the iceberg.digital research group and validant.ai platform