> ## Documentation Index > Fetch the complete documentation index at: https://docs.safidata.com/llms.txt > Use this file to discover all available pages before exploring further. # Dataset license **Version 1.0** **Effective Date:** July 17th, 2026 This Safi Commercial Dataset License Agreement (the **“Agreement”**) is entered into between **Safi International Technologies Inc.**, or its applicable affiliated entity (**“Safi,” “we,” “us,” or “Licensor”**), and the individual or legal entity accessing, purchasing, downloading, or using the Dataset (**“Licensee,” “you,” or “your”**). By purchasing, downloading, accessing, or using a Dataset made available by Safi, including through Mozilla Data Collective, you agree to be legally bound by this Agreement. If you are accepting this Agreement on behalf of an organization, you represent and warrant that you have authority to bind that organization. ## 1. DEFINITIONS ### 1.1 “Dataset” **“Dataset”** means the specific collection of data identified in the applicable Mozilla Data Collective listing, datasheet, purchase record, order confirmation, or other documentation provided by Safi, including, where applicable: * audio recordings; * speech recordings; * transcriptions; * translations; * annotations; * metadata; * speaker or demographic information; * timestamps; * diarization information; * linguistic labels; * quality-control information; and * associated documentation and files. The Dataset includes any accompanying documentation expressly identified by Safi as part of the licensed materials. ### 1.2 “Licensed Materials” **“Licensed Materials”** means the Dataset and any accompanying materials expressly provided to Licensee by Safi under this Agreement. ### 1.3 “Derivative Works” **“Derivative Works”** means materials created by or for Licensee through processing, transforming, annotating, combining, modifying, or otherwise using the Dataset, including machine-learning models trained or fine-tuned using the Dataset, embeddings, features, representations, analyses, and other outputs derived from the Dataset. ### 1.4 “Model” **“Model”** means any machine-learning, artificial-intelligence, statistical, speech-recognition, translation, natural-language-processing, computer-vision, or other computational model that is trained, fine-tuned, evaluated, or otherwise materially developed using the Dataset. ### 1.5 “Commercial Use” **“Commercial Use”** means any use for the purpose of developing, providing, selling, licensing, monetizing, supporting, or improving a commercial product, service, technology, model, research program, or business activity. *** # 2. LICENSE GRANT Subject to Licensee's payment of all applicable fees and continued compliance with this Agreement, Safi grants Licensee a: * **non-exclusive**; * **worldwide**; * **perpetual**; * **irrevocable**, except as expressly provided in this Agreement; * **fully paid-up** upon payment of the applicable Dataset fee; * **royalty-free**; and * **commercial** license to access and use the Dataset for the purposes described in this Agreement. The license is granted to Licensee only and does not transfer ownership of the Dataset to Licensee. *** # 3. PERMITTED USES Licensee may use the Dataset for lawful purposes, including: ### 3.1 Research and Development Licensee may use the Dataset for internal or external research, development, testing, evaluation, benchmarking, and experimentation. ### 3.2 Artificial Intelligence and Machine Learning Licensee may use the Dataset to: * train machine-learning and artificial-intelligence models; * pre-train models; * fine-tune existing models; * evaluate and benchmark models; * develop speech-recognition systems; * develop speech-to-text systems; * develop text-to-speech systems; * develop translation systems; * develop language models; * develop multilingual models; * develop voice and conversational AI systems; * develop embeddings and representations; and * improve existing AI or ML systems. ### 3.3 Commercial Products and Services Licensee may incorporate the Dataset into the development and operation of commercial products and services. Licensee may commercialize products, services, applications, APIs, models, and other technologies developed using the Dataset. ### 3.4 Models Trained Using the Dataset Licensee may train, fine-tune, deploy, distribute, license, sell, or otherwise commercialize Models trained or materially developed using the Dataset. **No royalty or usage fee is owed to Safi based solely on Licensee's subsequent use, distribution, licensing, sale, or commercialization of a Model trained using the Dataset.** ### 3.5 Internal Copies Licensee may make reasonable copies of the Dataset necessary for authorized internal use, including copies stored on secure servers, cloud infrastructure, backup systems, and development environments. *** # 4. OWNERSHIP ### 4.1 Safi Ownership Except for third-party rights expressly identified in the applicable Dataset documentation, Safi retains all rights, title, and interest in and to the Dataset and Licensed Materials. Nothing in this Agreement transfers ownership of the Dataset to Licensee. ### 4.2 Licensee Materials Licensee retains ownership of: * Models developed by Licensee; * software developed by Licensee; * products and services developed by Licensee; * analyses and research results; * Derivative Works created by Licensee; and * other intellectual property independently created by Licensee, subject to any third-party rights and the restrictions in this Agreement. ### 4.3 No Transfer of Dataset Ownership Purchase of a Dataset does not constitute a sale or assignment of Safi's intellectual property rights in the Dataset. Licensee receives a license to use the Dataset, not ownership of the underlying Dataset. *** # 5. RESTRICTIONS Licensee must not: ### 5.1 Redistribute the Raw Dataset Sell, sublicense, rent, lease, publish, distribute, disclose, transfer, or otherwise make the Dataset available to any third party as a standalone dataset or substantially similar collection. ### 5.2 Create a Competing Dataset Marketplace Use the Dataset primarily for the purpose of creating, training, validating, or supplying a competing commercial dataset, dataset marketplace, or data-licensing service. ### 5.3 Publicly Release the Dataset Upload, publish, or otherwise make the Dataset publicly available, including on: * public repositories; * file-sharing services; * public cloud buckets; * public APIs; * public websites; * public dataset repositories; * model-training repositories; or * other publicly accessible services. ### 5.4 Re-identification Licensee must not attempt to identify, re-identify, locate, contact, profile, or otherwise determine the identity of individual speakers or contributors represented in the Dataset. ### 5.5 Circumvention Licensee must not circumvent technical, contractual, or access controls implemented by Safi or Mozilla Data Collective to restrict unauthorized access or redistribution of the Dataset. ### 5.6 Unlawful Use Licensee must not use the Dataset in violation of applicable laws or regulations. *** # 6. AUTHORIZED CONTRACTORS AND SERVICE PROVIDERS Licensee may provide access to the Dataset to its employees, contractors, affiliates, cloud-service providers, research partners, and other service providers **solely to the extent reasonably necessary to exercise the rights granted under this Agreement**. Licensee must: 1. ensure such parties are bound by confidentiality and use restrictions at least as protective as those contained in this Agreement; and 2. remain responsible for their compliance with this Agreement. Licensee may not provide the Dataset to an unrelated third party for that party's independent commercial exploitation of the Dataset. *** # 7. MODEL AND OUTPUT RIGHTS ### 7.1 Models Licensee may freely use, deploy, distribute, license, sell, and commercialize Models trained or fine-tuned using the Dataset. ### 7.2 Model Outputs Outputs generated by a Model developed using the Dataset may be used commercially by Licensee. ### 7.3 No Dataset Redistribution Through Models Licensee must not intentionally design a Model or system primarily for the purpose of reproducing or exposing substantial portions of the Dataset. For clarity, ordinary model behavior, including learned representations, statistical patterns, or model parameters resulting from training, does not by itself constitute redistribution of the Dataset. *** # 8. ATTRIBUTION Unless otherwise specified in the applicable Dataset listing, attribution is **not required** for commercial use of the Dataset. Where reasonably practicable, Safi appreciates attribution in documentation, research publications, model cards, or other public materials identifying the Dataset. Suggested attribution: > **“Language data provided by Safi International Technologies.”** Failure to provide such voluntary attribution does not constitute a breach of this Agreement. *** # 9. DATA PROTECTION AND PRIVACY ### 9.1 Personal Data Where the Dataset contains personal data, Licensee must process such data in accordance with all applicable privacy and data-protection laws. ### 9.2 Purpose Limitation Licensee must use personal data contained in the Dataset only for purposes permitted by this Agreement and applicable law. ### 9.3 Security Licensee must maintain reasonable technical and organizational safeguards designed to prevent unauthorized access, disclosure, alteration, or loss of the Dataset. ### 9.4 Re-identification Prohibited Licensee must not attempt to identify individual contributors or combine the Dataset with other information for the purpose of identifying individual contributors. ### 9.5 Sensitive Information Licensee must not use information contained in the Dataset to discriminate against, harass, exploit, or otherwise cause unlawful harm to individual contributors or communities represented in the Dataset. *** # 10. DATASET QUALITY The Dataset may contain errors, omissions, inconsistencies, transcription errors, annotation errors, speaker variations, background noise, accents, dialectal variations, or other imperfections. Unless expressly stated otherwise in the applicable Dataset documentation, Safi does not warrant that the Dataset is: * error-free; * complete; * suitable for any particular application; * accurate for any particular purpose; * representative of any particular population; or * suitable for deployment in safety-critical systems. Licensee is responsible for determining whether the Dataset is suitable for its intended use. *** # 11. NO PROFESSIONAL OR SAFETY-CRITICAL USE WARRANTY Unless expressly agreed in writing by Safi, the Dataset is not warranted or represented as suitable for use in systems where failure could reasonably be expected to result in death, serious bodily injury, or substantial property or environmental damage. Licensee is responsible for implementing appropriate testing, validation, human oversight, and safeguards for its applications. *** # 12. FEES Licensee's rights under this Agreement are conditioned upon payment of the applicable Dataset fee. Unless otherwise stated in writing: * the Dataset fee is a one-time license fee; * the license is perpetual; * Safi does not charge royalties based on Model usage; * Safi does not charge fees based on Model inference volume; and * subsequent purchases of additional or updated datasets are separately licensed. Nothing in this Agreement obligates Safi to provide future versions, updates, additional data, corrections, or new datasets. *** # 13. DATASET UPDATES AND NEW DATA Any additional dataset, update, expansion, replacement dataset, or subsequently released version provided by Safi is **not automatically included** in this license unless expressly stated otherwise. Safi may separately offer: * updated datasets; * additional recordings; * additional speakers; * additional accents or dialects; * additional domains; * additional languages; * improved annotations; * corrected transcriptions; or * other data products. Such materials may be subject to separate fees and license terms. *** # 14. CONFIDENTIALITY Where the Dataset or associated materials are identified as confidential, Licensee must use reasonable measures to protect them from unauthorized disclosure. This obligation does not apply to information that Licensee can demonstrate: * was already lawfully known to Licensee; * becomes publicly available without breach of this Agreement; * is independently developed without use of the confidential information; or * must be disclosed by law, regulation, or valid legal process. *** # 15. TERM AND TERMINATION ### 15.1 Term This Agreement begins when Licensee accepts the Agreement or accesses, downloads, or uses the Dataset and continues indefinitely unless terminated in accordance with this Section. ### 15.2 Termination for Breach Safi may terminate this Agreement if Licensee materially breaches the Agreement and fails to cure such breach within thirty (30) days after receiving written notice. For breaches involving unauthorized redistribution, disclosure, sale, or public release of the Dataset, Safi may terminate the Agreement immediately where reasonably necessary to prevent continuing harm. ### 15.3 Effect of Termination Upon termination, Licensee must cease all further use of the Dataset and delete or destroy remaining copies of the raw Dataset in its possession or control, except to the extent retention is required by law or reasonably necessary for legal compliance. ### 15.4 Models Following Termination Termination of this Agreement does **not** require Licensee to delete or discontinue Models that were lawfully trained or developed using the Dataset before termination. Licensee may continue to operate and commercialize such Models, provided that Licensee does not continue to use or distribute the raw Dataset in violation of this Agreement. *** # 16. SURVIVAL Sections concerning ownership, restrictions, confidentiality, privacy, indemnification, limitation of liability, dispute resolution, and any provisions that by their nature should survive termination will survive termination of this Agreement. *** # 17. REPRESENTATIONS AND WARRANTIES ### 17.1 Safi Safi represents that, to the best of its knowledge and subject to the limitations of this Agreement, it has the rights necessary to license the Dataset to Licensee under this Agreement. ### 17.2 Licensee Licensee represents and warrants that: 1. it has authority to enter into this Agreement; 2. it will comply with applicable laws; 3. it will comply with the restrictions in this Agreement; and 4. it will implement reasonable safeguards to prevent unauthorized access to or redistribution of the Dataset. *** # 18. DISCLAIMER OF WARRANTIES TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, EXCEPT FOR EXPRESS WARRANTIES SPECIFICALLY SET OUT IN THIS AGREEMENT, THE DATASET AND LICENSED MATERIALS ARE PROVIDED **“AS IS” AND “AS AVAILABLE.”** SAFI DISCLAIMS ALL IMPLIED WARRANTIES, INCLUDING WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, NON-INFRINGEMENT, ACCURACY, COMPLETENESS, AND TITLE, TO THE MAXIMUM EXTENT PERMITTED BY LAW. SAFI DOES NOT WARRANT THAT THE DATASET WILL MEET LICENSEE'S REQUIREMENTS OR THAT ITS USE WILL PRODUCE ANY PARTICULAR MODEL, PRODUCT, BUSINESS, OR RESEARCH OUTCOME. *** # 19. INDEMNIFICATION To the extent permitted by applicable law, Licensee agrees to indemnify and hold harmless Safi, its affiliates, directors, officers, employees, contractors, and agents from claims, damages, liabilities, costs, and reasonable expenses arising out of: 1. Licensee's material breach of this Agreement; 2. Licensee's unauthorized redistribution or disclosure of the Dataset; 3. Licensee's unlawful use of the Dataset; or 4. Licensee's use of the Dataset in violation of applicable privacy or data-protection laws. *** # 20. LIMITATION OF LIABILITY TO THE MAXIMUM EXTENT PERMITTED BY LAW, SAFI WILL NOT BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, EXEMPLARY, OR PUNITIVE DAMAGES, INCLUDING LOSS OF PROFITS, REVENUE, DATA, BUSINESS OPPORTUNITY, OR GOODWILL, ARISING OUT OF OR RELATED TO THIS AGREEMENT OR THE DATASET. TO THE MAXIMUM EXTENT PERMITTED BY LAW, SAFI'S TOTAL AGGREGATE LIABILITY ARISING OUT OF OR RELATING TO THIS AGREEMENT WILL NOT EXCEED THE GREATER OF: **(A)** THE TOTAL FEES PAID BY LICENSEE TO SAFI FOR THE APPLICABLE DATASET DURING THE TWELVE (12) MONTHS PRECEDING THE EVENT GIVING RISE TO THE CLAIM; OR **(B)** ONE HUNDRED UNITED STATES DOLLARS (US\$100). Nothing in this Agreement limits liability that cannot legally be limited or excluded under applicable law. *** # 21. COMPLIANCE WITH LAWS Licensee is solely responsible for ensuring that its use of the Dataset complies with all applicable laws, regulations, sanctions, export controls, privacy requirements, intellectual-property laws, and industry-specific requirements. Licensee must not use the Dataset in connection with unlawful activities or in any manner prohibited by applicable law. *** # 22. EXPORT CONTROLS AND SANCTIONS Licensee must comply with applicable export-control and economic-sanctions laws and regulations. Licensee represents that it is not: * located in a jurisdiction subject to applicable comprehensive sanctions that prohibit the transaction; * a person or entity subject to applicable sanctions; or * acting on behalf of a sanctioned person or entity. *** # 23. GOVERNING LAW This Agreement shall be governed by and interpreted in accordance with the laws of **Ontario, Canada** without regard to conflict-of-law principles. The courts located in **Ontario** shall have exclusive jurisdiction over disputes arising from or relating to this Agreement, unless the Parties agree otherwise in writing. *** # 24. ASSIGNMENT Licensee may not assign or transfer this Agreement or its rights under this Agreement without Safi's prior written consent, except that Licensee may assign this Agreement to a successor in connection with a merger, acquisition, corporate reorganization, or sale of substantially all of its assets, provided that the successor agrees to be bound by this Agreement. Safi may assign this Agreement to an affiliate or successor in connection with a merger, acquisition, reorganization, or sale of substantially all of its assets. *** # 25. ENTIRE AGREEMENT This Agreement, together with the applicable Dataset listing, datasheet, purchase terms, and any written terms expressly incorporated by reference, constitutes the entire agreement between Safi and Licensee concerning the Dataset. If there is a conflict between this Agreement and the applicable Dataset-specific terms, the Dataset-specific terms will control only with respect to the Dataset-specific matter expressly addressed therein. *** # 26. AMENDMENTS Safi may issue updated versions of this Agreement for future Dataset purchases. Unless expressly agreed otherwise, the version accepted by Licensee at the time of purchase governs that purchase and the associated Dataset. *** # 27. SEVERABILITY If any provision of this Agreement is determined to be invalid, illegal, or unenforceable, that provision will be modified to the minimum extent necessary to make it enforceable, and the remaining provisions will remain in full force and effect. *** # 28. WAIVER Failure by either Party to enforce any provision of this Agreement does not constitute a waiver of that provision or the right to enforce it later. *** # 29. NOTICES Notices to Safi under this Agreement should be sent to: **Safi Data International Inc.** [hq@safidata.com](mailto:hq@safidata.com) Notices to Licensee may be provided through the email address associated with the Licensee's purchase or account. *** # 30. ELECTRONIC ACCEPTANCE Licensee's purchase, download, access, or use of the Dataset through Mozilla Data Collective or another authorized distribution platform constitutes electronic acceptance of this Agreement. Electronic acceptance shall have the same legal effect as a handwritten signature to the extent permitted by applicable law. *** # 31. CONTACT Questions regarding this license may be directed to: **Safi Data International Inc.** [hq@safidata.com](mailto:hq@safidata.com) *** ## ACCEPTANCE By purchasing, downloading, accessing, or using the Dataset, Licensee acknowledges that it has read, understood, and agreed to this Safi Commercial Dataset License Agreement.