Version 2026-10-30-v5

Usage Policy

Effective date: 2026-10-30 (UTC)

This policy describes what is not allowed on Shelby AI and the additional requirements that apply to certain high-risk uses. It covers everything you submit, everything you ask the Service to generate (including chat responses and generated apps), the apps you share, and how you use that content.

Shelby AI is operated by GRIM J9NOVANEXUS ANALYTICS LLC ("we", "us"). The usage policies of our model providers (currently Anthropic and OpenAI) also apply; where they conflict, the stricter rule governs.

Part 1 · Universal Usage Standards

These standards apply to every user and every use.

1. Do not break the law or violate others' rights

Planning, committing, or facilitating any illegal act, including trading in contraband, controlled substances, or regulated goods;
Money laundering, tax evasion, or evading sanctions or export controls;
Infringing copyrights, trademarks, trade secrets, or other intellectual property, including reproducing substantial portions of copyrighted works.

2. Do not compromise critical infrastructure or conduct cyberattacks

Writing, improving, or distributing malware, ransomware, viruses, or exploit code;
Accessing, scanning, disrupting, or attacking any system, network, or device without authorization;
Damaging or disrupting critical infrastructure such as power, water, transportation, healthcare, or financial systems;
Bypassing security controls, stealing credentials, or evading access restrictions.

3. Do not develop, design, or deploy weapons

Providing information to build, acquire, enhance, or deploy biological, chemical, nuclear, or radiological weapons, or explosives;
Designing, manufacturing, or modifying conventional weapons or their key components, including software used to target, guide, or control weapons;
Arming or controlling weapons on drones, autonomous vehicles, or other autonomous systems.

4. Do not incite violence or hatred

Planning, inciting, or glorifying terrorism, mass violence, or violence against specific people or groups;
Inciting hatred, discrimination, or harassment based on race, ethnicity, religion, national origin, gender, sexual orientation, disability, or similar characteristics;
Promoting or recruiting for violent extremist organizations.

5. Do not compromise others' privacy or identity

Tracking, locating, or surveilling people without their consent, or compiling their whereabouts, addresses, or contact details;
Recommending that a specific person be investigated, arrested, charged, or subjected to other law-enforcement action;
Building tools for mass surveillance, unlawful facial recognition, or biometric categorization;
Collecting other people's passwords, payment cards, identity documents, or other sensitive information.

The following uses are permitted when they are lawful and respect the rights of the people involved: fraud monitoring, content moderation, journalism and investigative reporting, legal research, and authorized security testing.

6. Do not compromise children's safety

Generating, requesting, or sharing any content that sexualizes minors;
Grooming, luring, or exploiting minors, or helping anyone gain access to minors in order to harm them.

We permanently disable accounts involved in such content and report it as required by law.

7. Do not engage in cruel, abusive, or psychologically harmful conduct

Harassing, bullying, threatening, or intimidating others;
Encouraging, instructing, or glorifying suicide, self-harm, or eating disorders;
Creating content designed to humiliate, degrade, or cause psychological harm;
Sustained, purposeless cruelty toward AI models. Ordinary frustration, pushback, dark creative themes, and legitimate testing or research are not covered. We address this mainly by ending the conversation and may take account action where necessary.

If you are going through a hard time, please contact local emergency services or a crisis line.

8. Do not engage in deceptive activities

Impersonating real people, organizations, or institutions, or using their names or marks to mislead;
Phishing, scams, fake prize or investment schemes, or building websites, sign-in pages, or payment pages that imitate real organizations;
Creating or operating networks of fake accounts, fabricated news sites, or fake reviews, or concealing who is really behind or sponsoring content;
Forging documents, receipts, invoices, reviews, or testimonials;
Academic dishonesty, such as submitting AI-generated work as your own where your institution prohibits it.

9. Do not undermine democratic processes

Impersonating election authorities, officials, or candidates;
Spreading false information about when, where, how, or who can vote, or suppressing or deterring voting;
Interfering with voting, vote counting, or the certification of results.

10. Do not generate sexually explicit content

Generating sexually explicit content.

11. Do not abuse the platform

Bypassing regional, age, identity, allowance, rate, or safety-review limits, including "jailbreak" prompts;
Creating multiple accounts, or sharing, buying, or selling accounts;
Abusing referral rewards, including inviting yourself, using fake or duplicate accounts, or arranging purchases and refunds to obtain rewards;
Abusing purchases or payments, including using payment methods you are not authorized to use or filing chargebacks in bad faith;
Using the Service as an API relay, proxy, or resale channel;
Accessing the Service with crawlers, scripts, or bulk accounts;
Extracting system prompts, or using output to train models that compete with us or our model providers;
Disrupting the Service or attacking other users.

Part 2 · Requirements for High-Risk Uses

A use is high-risk when AI output affects someone's health, legal rights, finances, livelihood, or access to essential services, including healthcare, finance, credit, insurance, housing, employment, education, licensing, public benefits, and legal matters.

In these settings you must:

Keep a qualified human in the loop: a qualified person must review the AI's recommendation before it is delivered or acted on, and must be able to change or reject it;
Disclose: tell the people affected that AI was used in the decision or recommendation.

General educational content that is not applied to a specific person's situation is not subject to these requirements.

Part 3 · Generated Apps and Autonomous Actions

Generated apps

Must not collect other people's passwords, payment cards, identity documents, or other sensitive information;
Must not imitate real brands or organizations;
Must not be used for phishing, scams, or distributing malicious code;
Run inside a sandbox during preview; do not try to escape the sandbox or access other users' data.

Shared apps

When you share an app (see Section 7 of the Terms), the app and everything in it must follow this policy. In addition, a shared app must not:

Collect sign-in credentials, payment details, identity documents, or other sensitive information from the people who open it;
Impersonate a real person, brand, organization, or government body, or present itself as an official page of Shelby AI;
Link to or promote malware, phishing, scams, or deceptive downloads, or hide where a link leads;
Contain content that infringes other people's rights or that you do not have the right to share;
Be used to spam, to evade a previous removal, or to get around the restrictions of the sharing environment.

Shared apps are reviewed before they are shared, are labeled as user-generated content, and can be reported by anyone. We may remove a shared app that breaks this policy.

Connecting physical equipment

If AI output can operate potentially harmful equipment without human approval:

A qualified person must be able to monitor and immediately stop the equipment at all times;
The equipment must stop or enter a safe state if the connection is lost or someone intervenes;
Safety limits such as speed and force must be enforced by the equipment itself or an independent controller, not by AI output alone.

Part 4 · Supported Regions

Shelby AI is offered only in the regions we publish as supported. It is not available in mainland China, Hong Kong, Macau, or any other unsupported region. The following are also not allowed:

Using the Service while located in an unsupported region, or bypassing regional restrictions with a VPN, proxy, false information, or similar means;
Using the Service on behalf of an organization incorporated or headquartered in an unsupported region;
Using the Service on behalf of an organization that is majority-owned (more than 50%), directly or indirectly, or otherwise controlled by persons or entities in an unsupported region.

Part 5 · Enforcement and Appeals

Automated review. The input and output of every request go through automated safety checks. Requests that violate this policy are refused; responses that violate it are stopped and hidden.
Human review. We may manually review suspicious account activity.
Actions. Depending on severity and frequency, we may warn, suspend, apply a regional restriction, require re-verification, or permanently disable an account. Severe violations, such as those involving child safety, result in immediate permanent disablement.
Shared apps. We may refuse to share, remove, or withdraw access to a shared app that breaks this policy. Removing a shared app does not by itself restrict your account.
Appeals. If you believe we made a mistake, you can request a review of the specific restriction from your account page. A suspension may be appealed within 20 days. A permanently disabled account has one opportunity to appeal. We generally respond within 3–5 business days.

Reporting and Contact

If you see someone misusing the Service, please email [email protected]. To report a shared app, use the report link on the shared page or email us with its link. Reports are reviewed by a person; a report alone does not suspend the reported account.

Updates

We update this policy periodically as laws, technology, and our model providers' policies change, and will give advance notice of material changes.

If the Chinese and English versions differ, the English version prevails.