legal: publish Neo Ar dataset licenses

legal/DATASET_LICENSE.md

@@ -0,0 +1,120 @@
+TokenAI Neo Ar Dataset License — Non-Commercial / No-Rebranding
+
+Copyright (c) 2026 TokenAI / Assem Sabry.
+
+TokenAI is a non-profit startup founded in 2025 by Assem Sabry, based in Alexandria, Egypt. Contact: info@tokenai.llc. Website: https://tokenai.llc.
+
+This license applies to the dataset published at https://huggingface.co/datasets/tokenaii/Neo-Ar-Dataset.
+
+## 1. Entire Agreement and acceptance
+
+This license and the files expressly referred to in it constitute the complete
+agreement between TokenAI and every person or organization that accesses,
+downloads, copies, reads, processes, transforms, uses, or trains on the Neo
+dataset or related materials. By doing any of those acts, the user
+acknowledges that they have read and accepted this agreement, its conditions,
+its attribution duties, its restrictions, and the consequences of breach.
+
+Earlier instructions, posts, announcements, examples, or informal statements
+do not override this license. Any exception or additional permission must be in
+a written instrument confirmed by an authorized TokenAI representative.
+Licenses governing external dependencies remain separate and continue to apply
+to those dependencies.
+
+## 2. Definitions
+
+**Commercial Use** means any use by or for a for-profit person or organization, or any use connected with a paid product, service, API, subscription, advertising, sponsorship, client work, employment, production business operation, resale, fundraising, investment, or business advantage.
+
+**Financial Benefit** includes revenue, fees, salary, commission, donations, sponsorship, investment value, cost savings, lead generation, competitive advantage, in-kind compensation, or any other direct or indirect economic value.
+
+**Derivative Model** means any model or model artifact materially derived from this dataset, including fine-tuned, merged, distilled, quantized, pruned, converted, adapted, LoRA/adapter, checkpoint, embedding, teacher/student, synthetic expansion, or behaviorally substituted model.
+
+**Official Dataset Name** means **TokenAI Neo Ar Decision Model Dataset**. The canonical dataset reference is `tokenaii/Neo-Ar-Dataset`.
+
+## 3. Limited permission
+
+Permission is granted only for personal, private, non-commercial educational, academic, and non-profit research use. No Commercial Use or Financial Benefit is permitted. This license does not grant permission to redistribute the data, operate a dataset service, or publish a Derivative Model.
+
+## 4. Mandatory identity and attribution
+
+Every permitted use must identify the data as **TokenAI Neo Ar Decision Model Dataset** and display:
+
+- Dataset: `tokenaii/Neo-Ar-Dataset` — https://huggingface.co/datasets/tokenaii/Neo-Ar-Dataset
+- Model/project: `tokenaii/Neo-Ar` — https://huggingface.co/tokenaii/Neo-Ar
+
+Every model, report, experiment, or derived artifact trained using this data must include:
+
+> This work was trained using the TokenAI Neo Ar Decision Model Dataset, `tokenaii/Neo-Ar-Dataset`, owned by TokenAI.
+
+Generator name, seed, manifest, split information, provenance notices, and this license must remain intact where they accompany the data.
+
+## 5. No redistribution, rebranding, or commercial use
+
+Without prior written permission from TokenAI, you may not:
+
+1. redistribute, rehost, mirror, sublicense, sell, publish, or make available the dataset or an unchanged or substantially equivalent copy;
+2. publish reconstructed copies, scraped mirrors, or replacement downloads;
+3. rename, rebrand, white-label, or publish the dataset under another identity;
+4. remove or obscure TokenAI, Assem Sabry, the Official Dataset Name, canonical references, provenance, or this license;
+5. use the dataset to train a commercial or monetized model, service, product, API, or workflow;
+6. use labels, checkpoints, embeddings, or synthetic expansions to evade these restrictions or to obtain Commercial Use or Financial Benefit;
+7. use `TokenAI`, `Neo`, or confusingly similar marks as another product or organization's identity.
+
+Publishing a model trained with this dataset never waives the attribution requirement.
+
+## 6. TokenAI and Neo trademark policy
+
+No trademark, trade-name, logo, domain-name, or endorsement rights are granted. Factual nominative reference is allowed only to identify the source and must not imply sponsorship or certification. No confusingly similar mark or domain may be registered or used. Written permission is required for co-branding, white-labeling, or branding beyond attribution.
+
+## 7. Written permission
+
+Request exceptions directly from TokenAI at **info@tokenai.llc** with the material, intended use, distribution plan, duration, and parties involved. Only a written instrument from an authorized TokenAI representative grants an exception.
+
+## 8. Notice-and-takedown
+
+TokenAI may send a written notice identifying the material or URL, the type of
+alleged violation, the specific license term or legal right that was breached,
+and a contact method or signature of the rights holder or its authorized
+representative.
+
+After receiving notice, the alleged violator must stop use, remove or disable
+access to the material within five business days, preserve relevant evidence
+and records, and confirm completion of the removal or access block to TokenAI.
+TokenAI may request assistance from a hosting platform or service provider. A
+counter-notice may explain the lawful basis for use and provide contact
+details, but it does not authorize continued Commercial Use or suspend
+termination. Urgent or repeated breaches may require immediate removal and
+termination. This process does not limit any other legal remedy.
+
+## 9. Contributions and future CLA
+
+No contribution is accepted unless expressly invited in writing. Contributors must sign a CLA, confirm ownership or authorization, disclose third-party material, and confirm the contribution contains no secrets, unlawful personal data, or incompatible restrictions. The CLA must grant TokenAI a perpetual, worldwide, irrevocable, royalty-free right to use, modify, publish, sublicense, and relicense the contribution. Contributions do not grant redistribution rights to this dataset.
+
+## 10. Third-party material and dependency review
+
+Third-party datasets, software, tokenizer assets, fonts, and other components retain their own licenses. Users must review the dependency and provenance manifest, preserve upstream notices, comply with each license, and verify compatibility before redistribution or deployment. No third-party material is relicensed by this document.
+
+## 11. Patents
+
+No patent license is granted, whether expressly, by implication, estoppel, or otherwise. All patent rights are reserved.
+
+## 12. Termination
+
+Any breach automatically terminates permission. All use and copies must stop and be removed after termination. TokenAI reserves all legal remedies.
+
+## 13. Severability
+
+An invalid provision will be limited or severed only as necessary; the rest of the license remains effective.
+
+## 14. Amendments
+
+TokenAI may amend the terms that accompany future releases, future dataset
+records, future code, or future publications. Materials already received remain
+governed by the terms supplied with them unless a written permission or written
+amendment confirmed by the affected user states otherwise. No amendment
+retroactively expands a recipient's rights or removes an existing restriction
+without that recipient's written agreement.
+
+## 15. Disclaimer
+
+THE DATASET IS PROVIDED "AS IS" WITHOUT WARRANTY OF ANY KIND. THIS CUSTOM LICENSE IS NOT LEGAL ADVICE AND SHOULD BE REVIEWED BY QUALIFIED COUNSEL BEFORE ENFORCEMENT.

legal/NOTICE.md

@@ -0,0 +1,14 @@
+# TokenAI Neo Ar Dataset Notice
+
+Official model name: **TokenAI Neo Ar — Arabic Decision Model**
+
+Canonical model: `tokenaii/Neo-Ar`
+
+Training dataset: `tokenaii/Neo-Ar-Dataset`
+
+TokenAI is a non-profit startup founded in 2025 by Assem Sabry, based in Alexandria, Egypt.
+Contact: info@tokenai.llc · https://tokenai.llc
+
+This model may not be rebranded, renamed, white-labeled, redistributed, or
+used commercially. Any permitted use must preserve the full official name and
+state that the model was trained using `tokenaii/Neo-Ar-Dataset`.