Our IP Philosophy
Innovation Protection
NexusAI invests heavily in research and development. Our proprietary architectures, training pipelines, and inference optimizations are protected through a comprehensive portfolio of patents and copyrights to ensure sustainable innovation.
Responsible Openness
We believe in strategic open-source contributions that accelerate the AI ecosystem. Base models, research tools, and utility libraries are released under permissive licenses, while production-ready enterprise features remain commercially licensed.
Data & Model Sovereignty
Customer data, fine-tuned models, and custom prompts remain exclusively owned by the client. NexusAI never trains on proprietary client data without explicit, revocable consent. Clear data separation is architecturally enforced.
Patent Portfolio
| Patent / Application | Technology Area | Status | Year |
|---|---|---|---|
| US11,294,882 B2 | Dynamic Inference Routing | Granted | 2023 |
| US11,452,109 B1 | Adaptive Context Window Optimization | Granted | 2024 |
| US18/772,410 | Federated Learning Coordination Protocol | Pending | 2024 |
| EP4088291 A1 | Real-Time Adversarial Defense Layer | Granted | 2025 |
| US18/901,225 | Multi-Modal Token Compression | Pending | 2025 |
A complete registry of granted patents, published applications, and trademarks is available in our official IP disclosures repository.
Licensing & Open Source Policy
Open Source Models & Libraries
NexusAI Core Utilities, NexusTokenizer, and NexusEval are released under the Apache 2.0 License. These components may be used, modified, and distributed commercially without royalty obligations, provided original copyright notices are preserved.
Base foundation models (Nexus-7B, Nexus-13B) are available under the Nexus Open Research License (NORL-1.0), permitting academic, non-profit, and commercial experimentation with usage caps of 100M tokens/day for free tiers.
Commercial & Enterprise Licensing
Production deployments, fine-tuning APIs, and on-premise installations require a commercial license. Tiered licensing is based on throughput (tokens/requests), model variants, and deployment environments (cloud, edge, air-gapped).
Enterprise agreements include indemnification, SLA guarantees, and optional IP co-ownership structures for jointly developed custom architectures.
Third-Party Components
NexusAI integrates several third-party open-source libraries (e.g., PyTorch, NumPy, FastAPI, ONNX Runtime). All dependencies are tracked via SBOM (Software Bill of Materials) and comply with their respective MIT, BSD, or Apache licenses.
Model weights trained on public datasets comply with dataset-specific attribution and usage restrictions. We maintain a transparent provenance tracker for all training data sources.
Trademarks & Brand Usage
"NexusAI", "Nexus Engine", and associated logos are registered trademarks. Partners may use approved logos under our partner program guidelines. Unauthorized reproduction, derivative branding, or misleading attribution is strictly prohibited.
For brand usage requests, consult our Brand Guidelines or contact the legal team.
Data Ownership & Compliance
Customer Data Rights
- 100% data isolation per tenant
- Zero retention beyond processing unless explicitly configured
- Exportable model checkpoints & fine-tuned weights
- Right to deletion & data portability (GDPR/CCPA aligned)
AI Safety & Governance
- Red-teaming & bias auditing documented per release
- Model cards & risk assessments publicly available
- Compliance with NIST AI RMF, EU AI Act (Class 2/3)
- Transparent watermarking for synthetic outputs
IP Licensing & Legal Inquiries
For partnership discussions, custom licensing, trademark clearance, or legal reviews, our IP and compliance team is available to assist.