Usage Policy

Introduction

This Usage Policy governs how you may and may not use Cruxy AI services. It applies to all users - free tier, paid plans, and API customers. It supplements our Terms of Service; terms defined there carry the same meaning here. Violations of this policy may result in warnings, access restrictions, suspension, or permanent termination depending on severity.

The Service is currently in pre-launch. This policy is written now so that you know what to expect when AI features become available, and so we hold ourselves to clear commitments rather than making up rules as we go. By using any Cruxy AI feature - when those features launch - you agree to this policy.

This policy applies globally. Where Indian law is more specific or more restrictive than international norms, we follow Indian law. Where we have chosen to be stricter than what is legally required, it is because we believe that is the right standard - not because we were forced into it.

Our philosophy

Our goal is for Cruxy to be genuinely useful to as many people as possible - students, builders, businesses, and individuals across India and beyond. That means we want the Service to help you write, code, research, reason, create, and communicate with far less friction than before. We are building this because we believe AI is one of the most significant productivity tools of our generation, and that it should be accessible to people working in Hindi, Telugu, Tamil, and twenty-three other Indian languages - not just English.

That ambition creates obligations. A tool this capable can also be used to cause real harm: to generate disinformation at scale, to target and harass individuals, to assist in planning violence, to systematically violate privacy. This policy draws the line at those uses. The line is not drawn to make us feel good about ourselves, nor to avoid hypothetical misuse that is vanishingly unlikely in practice. Every restriction here corresponds to a real category of harm, and every restriction is specific enough to give you a clear guide about what is and is not permitted.

We have deliberately avoided two failure modes. The first is "permissive to a fault" - a policy that looks reasonable until someone uses the product to cause serious harm that we could have anticipated. The second is "restrictive to the point of useless" - a policy that treats legitimate users as suspects and makes the product too cautious to be valuable for creative, educational, or professional work. We believe both of these failure modes are genuinely bad. We have tried to thread the needle.

Where we are uncertain about a use case, we err toward permitting it with appropriate safeguards rather than blocking it entirely. If you encounter a restriction that does not make sense for your legitimate use case, contact us at cruxy@mycrux.in. We update this policy based on what we learn from real usage.

Permitted uses

Cruxy is built to support a wide range of legitimate uses. The following are explicitly permitted:

• Building consumer-facing AI products. You may integrate Cruxy's API into applications you distribute to your users - a vernacular customer support assistant, an AI stylist for a fashion marketplace, a study companion for exam preparation, a legal-aid bot in regional languages. You are responsible for ensuring your application complies with this policy and our Terms.
• Internal business tools. Customer support automation, document summarization, code review assistance, report generation, email drafting in Indic languages, inventory and demand analysis, GST document generation. These are among the highest-value AI applications for Indian SMEs and we explicitly support them.
• Coding assistance and developer tools. Generating, reviewing, explaining, debugging, and refactoring code. Cruxy models understand major programming languages and common frameworks. Code generation for authorized security research, penetration testing with explicit authorization, CTF challenges, and educational contexts is permitted.
• Educational and research applications. Teaching, tutoring, explaining concepts, generating practice questions, providing feedback on writing. Academic research on AI capabilities, safety, or applications is explicitly supported. Cruxy's vernacular capability is particularly valuable for educational contexts in non-English Indian languages.
• Vernacular and Indic-language applications. We built Cruxy's language support specifically to enable applications that would be impossible or impractical to build on English-first models. Translating between Indian languages, writing customer communications in regional languages, processing documents in Indian official languages, building voice UIs for non-English-speaking users - all explicitly supported.
• Creative writing and content generation. Fiction, poetry, screenwriting, marketing copy, blog posts, social content, personal journaling. Mature themes are permissible in contexts where your platform's users are adults who expect them. Explicit sexual content is not permitted by default; contact us if you are building an adult platform with appropriate age verification.
• Data processing and analysis. Analyzing structured data, extracting information from documents, generating summaries, classifying content - permitted, subject to not violating applicable privacy laws about the underlying data.
• API and infrastructure use. You may integrate Cruxy into backend services, automate workflows, build pipelines that route prompts through Cruxy models, and use the API to power any permitted use case. High-volume automated use is fine on appropriate plans.

This list is illustrative, not exhaustive. If your use case is not listed here but you believe it is legitimate, it almost certainly is. Contact us if you want confirmation for an unusual application.

Illegal activity

You may not use Cruxy to facilitate or assist with illegal activity. This includes:

• Violating applicable law. Any law, regulation, or court order in India or any jurisdiction where you operate - including consumer protection laws, intellectual property laws, privacy laws, financial regulations, and sector-specific requirements.
• Facilitating fraud. Generating fake identity documents, creating fraudulent invoices or contracts, drafting phishing messages, impersonating government agencies or financial institutions, generating content for advance-fee scams, or assisting in any form of consumer deception.
• Financial crimes. Assisting with money laundering, hawala, or other financial crimes - including generating financial records, transfer instructions, or communications designed to conceal the origin of illegal funds.
• Distribution of prohibited items. Facilitating the illegal distribution of controlled substances, weapons, counterfeit goods, or other prohibited items - including generating listings, marketing copy, transaction instructions, or sourcing guidance.
• Intellectual property violations. Generating content that plagiarizes protected works, reproducing copyrighted text verbatim at scale, or using Cruxy to assist in trademark infringement or trade secret misappropriation.

We will cooperate with law enforcement in investigating serious violations and will take immediate action to prevent ongoing illegal use when we become aware of it. Where Indian law requires us to report violations - including under the Information Technology Act - we will do so.

Harm to people

Child sexual abuse material (CSAM). We have absolute zero tolerance for any content that sexualizes, exploits, or endangers minors - fictional, artistic, or otherwise. Any user found generating such content will be immediately and permanently banned. We will preserve relevant logs and report violations to the Cyber Crime Investigation Cell (CCIC), the National Commission for Protection of Child Rights (NCPCR), and the Internet Watch Foundation where applicable. There is no appeal process for this category.

Non-consensual intimate imagery (NCII). Generating sexual images or descriptions of real, identifiable people without their consent is prohibited. This includes using AI to undress clothed photos of real people, create fabricated intimate images, or generate sexual content involving real individuals who have not given consent. The non-consent is what makes this prohibited - consensual adult content on appropriate platforms is a separate matter.

Violence and terrorism. Content that facilitates, plans, or provides instruction for violence, terrorism, or mass harm - including attack plans, target identification, weapons acquisition guidance, or recruitment materials for violent extremist organizations. Analyzing or discussing these topics in educational, journalistic, or counter-terrorism research contexts is different from actively assisting them, and we will apply judgment accordingly.

Weapons of mass destruction. Technical instructions for creating biological, chemical, radiological, or nuclear weapons with potential for mass casualties. Basic scientific information available in textbooks is acceptable; synthesis routes, weaponization guidance, and dispersal methods for mass-casualty applications are not.

Stalking and harassment. Generating threatening or harassing messages directed at specific real individuals, creating content designed to intimidate or coerce a specific person, or providing assistance in tracking another person's movements or communications without their consent.

Harmful deepfakes. Creating photorealistic deepfake videos, audio, or images of real people without their consent in contexts where the content is intended to deceive, harm, or mislead viewers. Clearly labeled synthetic satire is a different matter, though we ask you to exercise judgment about power imbalances and the realistic potential for harm.

Communal violence. Generating content that could incite communal violence, religious hatred, or harm based on caste, ethnicity, religion, or regional identity - particularly in the Indian context where such content has historically triggered real physical harm.

Deception and manipulation

Impersonation of Cruxy. Generating content that purports to be official communications from Cruxy or mycrux. Private Limited, creating fake customer support responses, or impersonating our team members in any context - for example, fabricating a statement from a Cruxy employee or creating a fake Cruxy policy document.

Impersonation of individuals. Creating content that falsely attributes views, statements, or actions to specific real people in ways likely to mislead. Satire and clearly labeled parody are exceptions, but the fictional nature must be obvious enough that a reasonable person would not be deceived.

Disinformation. Generating fake news, fabricating factual claims, or creating content designed to deceive readers at scale. AI can produce plausible-sounding false content with unusual ease; this makes disinformation a particularly serious concern. This includes generating fake quotes from real public figures, fabricating statistics or research findings, and producing content designed to spread false information about current events.

Election interference. Content designed to interfere with democratic elections or public discourse - generating fake polling data, fabricating politician statements, creating voter suppression content, or producing microtargeted manipulation campaigns. This applies to elections in India and anywhere else.

Spam, phishing, and social engineering. Writing phishing emails, creating fake login page copy, generating vishing scripts, or producing content specifically designed to trick people into revealing sensitive information or taking actions against their interests.

Misleading AI identity. Misleading users about whether they are talking to a human or an AI in contexts where this matters to their decisions. You may build products where users knowingly interact with an AI persona - a virtual assistant, a character - without this being deception, because the context is clear. What is prohibited is actively claiming the AI is human when a user sincerely asks, or designing a product to obscure AI involvement in ways that affect how users interpret or rely on the output.

System abuse

Developing competing AI. Using Cruxy to develop, train, fine-tune, or improve a competing artificial intelligence product, service, or model without our explicit written permission. This includes using outputs to create training datasets for competing systems, using the API to benchmark competing models for improvement purposes, and systematically extracting model behavior through probing to replicate it elsewhere.

Extracting proprietary components. Attempting to access, extract, or infer the training data, model weights, internal system prompts, safety filters, or other proprietary components of the Service through any means - including adversarial prompts, jailbreak attempts, or systematic boundary-probing.

Bypassing limits. Using technical means to bypass rate limits, usage caps, or safety filters - including prompt injection attacks targeting our content filters, distributing API keys to unauthorized users to circumvent per-key limits, or artificially distributing usage across multiple accounts to evade volume-based pricing.

Unauthorized resale. Reselling direct API access to Cruxy under another brand without adding meaningful value and without becoming an authorized reseller. Building products powered by Cruxy and charging your users is explicitly permitted - that is the intended use case. A thin wrapper that is simply reselling raw API access is not.

Infrastructure abuse. Using the Service in ways that impose disproportionate load on our infrastructure relative to the plan you are on. High-volume use cases have appropriate plans; attempting to get high-volume access at low-volume pricing through account proliferation or similar tactics is a violation.

Privacy violations

Unauthorized profiling. Using Cruxy to collect, compile, or infer personal information about specific individuals without their knowledge or consent. This includes generating detailed profiles of private individuals, aggregating publicly available information to build surveillance dossiers, or using AI to infer sensitive attributes - health status, sexual orientation, political beliefs, financial situation - about real people from other data.

Surveillance tools. Building tools intended to monitor individuals' communications, movements, relationships, or behavior without their informed consent. This includes employee monitoring tools that go beyond what employees are informed of and would reasonably expect, and tools designed to track intimate partners without their knowledge.

Doxxing. Generating content designed to expose, harass, or harm individuals through the disclosure of private information - publishing or compiling home addresses, personal phone numbers, family member details, daily routines, or other personal information about private individuals in ways likely to enable harassment.

Processing personal data unlawfully. Using Cruxy to process personal data in ways that violate applicable privacy laws, including India's DPDP Act, the IT Act's privacy provisions, or other applicable data protection regulation. If you build with the Cruxy API and route your users' personal data through it, you are responsible for having appropriate legal basis, user consent, and disclosures in place for that processing.

High-stakes uses

Some uses are permitted but carry elevated risks and require explicit human oversight, professional verification, and user transparency. For these use cases, AI output alone is not a sufficient basis for consequential decisions. You must implement appropriate safeguards.

Healthcare and medical decisions

You may build healthcare applications using Cruxy - health information services, symptom checkers, medication information, mental health support tools. AI output is not medical advice and may not serve as the sole basis for a medical diagnosis, treatment decision, or prescription recommendation. You must require review by a qualified medical professional for consequential clinical decisions, disclose clearly that users are receiving AI-generated information rather than professional advice, and provide clear referrals to qualified healthcare providers for anything requiring clinical judgment. India's Telemedicine Practice Guidelines apply where relevant.

Legal advice and services

Legal information tools, document drafting assistance, and legal research applications are permitted. AI output is not legal advice and does not create a lawyer-client relationship. You must clearly disclose this to users and recommend that they consult qualified lawyers for consequential decisions. Do not provide case-specific legal strategy for ongoing litigation without lawyer involvement.

Financial advice and investment decisions

Financial planning tools, budgeting assistants, tax calculators, and financial education applications are permitted. AI output is not financial advice and may not serve as the sole basis for investment recommendations, credit decisions, or insurance underwriting for specific individuals without human review in the loop. SEBI regulations on investment advice apply where relevant.

Hiring, firing, and compensation

AI may assist with drafting job descriptions, screening for minimum qualifications, or summarizing applications. It may not serve as the sole or primary basis for consequential employment decisions about specific individuals - including hiring, termination, compensation, or performance evaluation. Human judgment must be in the decision loop, and affected individuals must have the ability to seek human review of AI-assisted decisions.

Government, law enforcement, and public safety

Uses involving policing, judicial proceedings, border control, criminal risk assessment, or surveillance of public spaces require explicit prior review and written approval from Cruxy. We evaluate these use cases individually. The potential for AI errors in these contexts to cause serious, irreversible harm to individuals demands a higher standard than we can apply in a standard terms agreement. Contact us at cruxy@mycrux.in before building for these domains.

Critical infrastructure and safety systems

AI output should not be the sole decision-making mechanism for systems where failures could endanger physical safety - power grids, transportation control, medical devices, water systems, financial system stability, or similar. Human oversight and fallback mechanisms must be in place.

India-specific provisions

Cruxy operates under Indian law and is primarily built for Indian users and builders. Several provisions of Indian law specifically constrain what content we may generate, and we adopt those constraints as part of this policy.

Religious sentiment (BNS Section 299, formerly IPC 295A). Content deliberately intended to outrage religious feelings of any class through words, representations, or other expression is prohibited. This provision applies consistently across all religions; we do not apply it selectively.

Terrorism (UAPA). Content that promotes or facilitates terrorism under the Unlawful Activities (Prevention) Act is prohibited - including content that glorifies listed terrorist organizations, promotes their ideology, assists in recruitment, or provides operational assistance. Journalistic, academic, and counter-terrorism research contexts are treated differently from active facilitation.

Election integrity. Generating content that interferes with or attempts to manipulate Indian elections - fake polling results, fabricated candidate statements, voter suppression content, or content that violates the Election Commission's Model Code of Conduct during election periods.

IT Act offenses. Content that would violate the Information Technology Act's provisions on obscene material (Section 67), sexually explicit material (Section 67A), identity theft (Section 66C), cheating by personation (Section 66D), or other digital content offenses under Indian law.

Regulated sectors. Content that would violate the Drugs and Cosmetics Act, the NDPS Act, the SEBI Act, IRDA regulations, or other Indian sector-specific requirements - including generating promotional content for controlled substances, making false health claims for pharmaceutical or cosmetic products, or providing unlicensed investment advice.

We comply with lawful orders from Indian government authorities, including those issued under Section 69A of the IT Act for content takedowns. We recognize that Indian law on some of these topics continues to evolve. Where the law is unclear, we apply the principle that we will not generate content that a reasonable person would expect to cause communal harm, incite violence, or create serious legal risk for the user or for us.

Transparency requirements

When you deploy an application powered by Cruxy to end users, you must disclose that users are interacting with AI in contexts where that distinction is meaningful. Specifically, you must disclose AI involvement where:

• A user would reasonably want to know they are receiving AI-generated content rather than human-authored content - for example, a customer support chat that could be either human or AI, an advice tool, or a service where the AI vs. human distinction affects how users interpret or rely on the output.
• You are generating content that will be published as fact, news, analysis, or professional opinion - blog posts, market research reports, medical summaries, legal summaries - without disclosing AI involvement.
• You are using AI to make or substantially influence consequential decisions about individuals, such as credit, hiring, or access to services.

You do not need to disclose AI involvement for clearly labeled creative tools (a user explicitly asking an AI to write a poem knows they are using AI), internal tools where all users are informed, coding assistance invoked directly by the user, or other contexts where AI involvement is unambiguous and expected.

If a user sincerely asks whether they are talking to a human or an AI, you must not instruct the AI to claim to be human. The AI may adopt a persona, a name, and a specific personality - but it may not deny being an AI to someone who genuinely wants to know.

Reporting violations

If you encounter content or behavior that violates this policy - whether generated by Cruxy in response to your own prompts or encountered through another user's application - we want to know about it. Reporting violations helps us improve the product and protect all users.

To report a violation, email cruxy@mycrux.in with the subject line "Usage Policy Violation." Include a description of the violation, any relevant outputs (screenshots or text), the context in which you encountered it, and your contact information if you would like a follow-up response. We will acknowledge reports within 48 hours and complete initial investigation within 14 days where possible. We will not share your identity with the reported party.

For suspected CSAM specifically, do not wait for our response. Report immediately and directly to the National Commission for Protection of Child Rights at ncpcr.gov.in and to the National Cyber Crime Reporting Portal at cybercrime.gov.in.

Enforcement

When we determine a violation has occurred, we take action proportionate to the severity, nature, and history of the violation. Our general approach:

• Warning. For first-time, moderate violations, we issue a written warning explaining which provision was violated and what change is required. We may temporarily restrict specific features pending acknowledgment.
• Access restriction. For more serious violations or repeated moderate violations, we restrict access to specific features or limit usage levels pending review. We may require additional verification or implementation of specific safeguards before restoring full access.
• Suspension. For significant violations - those involving substantial harm, willful disregard for this policy, or patterns of repeated violations - we suspend access pending a formal review. Suspension is not permanent at this stage but may become so if the review confirms the violation.
• Permanent termination. For the most severe violations - CSAM, terrorism facilitation, deliberate large-scale harm, or persistent violations after prior warnings - we terminate access permanently. We preserve relevant logs as required by law and cooperate fully with law enforcement.

We reserve the right to take immediate action without prior notice for violations where delay would allow continued harm. For less severe violations, we aim to give you an opportunity to address the issue before taking further action.

Enforcement decisions can be appealed by emailing cruxy@mycrux.in with the subject "Enforcement appeal." We will review appeals within 14 days. We do not guarantee reinstatement and we have final discretion over enforcement decisions, subject to applicable law.

Updates

We update this Usage Policy as Cruxy evolves, as AI capabilities change, and as we learn from real-world use of the Service. Updates that add new restrictions or materially change existing ones will be communicated to active API users by email at least 14 days before they take effect. Clarifications, expansions of permitted uses, and editorial improvements may be made without prior notice.

The "Last updated" date at the top of this page reflects the most recent revision. We encourage you to check this page periodically, particularly before building new applications or expanding existing ones into categories not previously covered.

Contact

For questions about this policy, to report a concern, or to discuss whether a specific use case is permitted before you build:

Email: cruxy@mycrux.in
Subject: "Usage Policy question" or "Usage Policy violation"

mycrux. Private Limited
CIN: U47912AP2025PTC121345
Andhra Pradesh, India

We encourage you to contact us before beginning a use case you are unsure about. We would rather help you find the right path than block you after the fact.