The provided document outlines the Muse Model License (version: Muse Spark Open), established by Meta Platforms, Inc. on October 15, 2026, governing the use of the "Muse Spark Open" artificial intelligence model. This legal framework grants a non-exclusive, royalty-free license to access, modify, distribute, and commercialize the model, subject to mandatory compliance with an integrated Usage Policy (Exhibit A).

Core Legal and Operational Obligations

1. Attribution Requirement: Licensees must explicitly state "Built with Muse Spark Open" in the user interfaces, documentation, websites, or settings menus of any products or services utilizing the software or its derivatives.
2. Model as a Service (MaaS) Restriction: A specific commercial constraint is imposed on entities providing third-party access to AI models via API or similar interfaces. If the aggregate gross revenue of the Licensee and its affiliates exceeds $20,000,000 USD within a rolling 12-month period, the entity must notify Meta and execute a separate, formal written agreement. This requirement excludes internal research, non-MaaS commercial activity, and the integration of the model as a backend component of a product that is not itself a MaaS offering.
3. Intellectual Property: Meta expressly disclaims ownership over any outputs generated by the software.
4. Liability Disclaimer: The software is provided "AS IS," and the Licensee assumes full responsibility for the safety, deployment, and operational consequences of their derivatives. Meta disclaims all warranties (merchantability, fitness for a particular purpose, non-infringement) and liability for damages arising from the use of the model.

Usage Policy (Exhibit A)

The Usage Policy mandates ethical deployment and prohibits the use of the model for activities that infringe on rights or violate laws. Prohibited domains include:

• Illegal Activities: Promotion of violence, terrorism, exploitation of minors, human trafficking, or non-consensual sexual content.
• Discrimination and Harassment: Facilitating bullying, discriminatory practices in housing or employment, or the unauthorized practice of regulated professions (legal, medical, or financial).
• Privacy and Malice: Inference or collection of private sensitive data, the generation of malicious code/malware, or intentional circumvention of safety protocols.
• Harmful Physical Impact: Use related to military or warfare applications, the development of illegal weapons/drugs, or the management of critical infrastructure.
• Deception and Disinformation: Generating fraudulent, defamatory, or spam content; impersonation; or misleading users regarding the human-generated nature of AI outputs.
• Safety Disclosure: Users are obligated to communicate known limitations or dangers of the AI system to their own end-users.

Technical and Definition Scope

• Derivatives: Defined as any AI model that is "created, trained, modified (including fine-tuned), distilled, or otherwise improved using the Software or any outputs or results thereof."
• Jurisdiction: Meta’s legal entity and applicable privacy policies vary based on the Licensee's location, specifically distinguishing between the European region and the rest of the world.