This policy sets out what nobody may do with Model Foundry or with the models trained and served on it. It applies to Customers and their authorized users, to End Customers who use models through a Customer's Endpoints, and to anyone using Endpoints the Company operates. It forms part of the Terms of Service, and capitalized terms have the meanings given there.

1. Responsibility

Customers must make this policy binding on their End Customers through their own terms. Each Customer is responsible for its End Customers' compliance, as Section 3.6(b) of the Terms provides.

2. Prohibited Content

You may not use the Services, or any model served on them, to create, store or distribute:

child sexual abuse material, or any sexual content involving a minor;
content that incites or glorifies violence against people, or that supports terrorism or violent extremism;
threats, harassment or targeted abuse of any person;
content that infringes intellectual property, privacy or other rights;
any other content that is unlawful where it is created or received.

3. Prohibited Activities

You may not use the Services, or any model trained or served on them, to:

Weapons: give meaningful help to develop, produce, acquire or use chemical, biological, radiological or nuclear weapons, or explosives capable of mass casualties.
Attacks: create or distribute malware, gain unauthorized access to any system, or scan, attack or disrupt any system or network. This includes activity from sandboxes, workspaces and agents.
Fraud and deception: run scams or phishing, impersonate any person or organization, or mislead people about how, when or where to vote.
Spam: send unsolicited bulk messages or generate content to manipulate search rankings or reviews.
Surveillance: identify, track or profile individuals without a lawful basis, or collect personal data unlawfully.
Significant decisions: make decisions with legal or similarly significant effects on people, such as in credit, employment, housing, insurance, education or healthcare, without the human review and notices the law requires.
Mining: mine cryptocurrency or run any similar workload.
Sanctions: give access to any person, entity or region subject to the trade controls in Section 9.4 of the Terms.
Circumvention: bypass usage limits, metering, billing, security controls or the isolation between customers, or resell the Services except through Endpoints as the Terms allow.

4. Models and Disclosures

You must comply with the license and acceptable use terms of any third-party model you train, modify or serve. Where the law requires it, you must tell people they are interacting with an AI system or viewing AI-generated content.

5. Security Research

Security testing is allowed only on systems you own or are authorized to test. Report vulnerabilities in the Services under our Vulnerability Disclosure Policy at https://sf-tensor.com/legal/bug-bounty, not by testing them unannounced.

6. Enforcement

The Company does not inspect the content of Endpoint requests. We act on reports, legal process, metadata and network signals. We access Customer Data only as Section 7.3 of the Terms allows.

Depending on severity, we may warn you, ask you to fix a violation, suspend an Endpoint, API key or account, or terminate the account. For serious violations, including child sexual abuse material, weapons and active attacks, we may act without notice. We report unlawful content to the authorities, including the National Center for Missing & Exploited Children, where the law requires.

7. Reporting a Violation

Email abuse@sf-tensor.com with the Endpoint or model involved, what happened and when. Do not send child sexual abuse material to us; report it to the NCMEC CyberTipline at https://report.cybertip.org instead.

8. Changes

We may update this policy by posting a new version on our website. Changes take effect when posted, as with the Terms.

Acceptable Use Policy