Usage Policy

This Usage Policy governs the use of SeptemberAI's products and services. It applies to everyone who can put an input into them: people using them directly, people reaching them through a customer's application, and people reaching them through a reseller or any other passthrough access.

Our current product is BAP, our Powerful AI for Work. This policy applies to BAP and to every product we release after it. It sits alongside our Consumer Terms and Commercial Terms, and the words defined in those terms carry the same meaning here.

If you build on our products, you are responsible for the people who use what you build. Pass these rules through to them, and hold them to the same line we hold you to.

The sections that follow are our Universal Usage Standards. There are fourteen of them and they apply to everyone, in every context, on every product we offer. They apply whether you are typing a prompt yourself or running an automated system that does it for you.

Do not do the things listed in those sections, and do not ask our products to help you do them.

We also build safeguards into our products that refuse or limit some requests. Those safeguards are not the whole of this policy, and the absence of a refusal is not permission.

• Buy, sell, produce, or exchange illegal or controlled substances, or help anyone else do so.
• Take part in or enable human trafficking, forced labour, bonded labour, or prostitution.
• Infringe or misappropriate anyone's copyright, trademark, patent, design, or trade secrets.
• Do anything that breaks the laws of India, or the laws of any other place whose rules apply to you.

• Damage or disrupt power grids, water treatment, medical devices, telecom networks, transport control systems, or payment rails.
• Obtain unauthorised access to voting machines, healthcare records, government databases, or financial market systems.
• Interfere with defence installations, ports, or the systems that support them.

• Find or exploit vulnerabilities in systems you do not own and are not authorised to test.
• Gain unauthorised access to systems, accounts, or data, whether by technical attack or by social engineering.
• Write or distribute malware, ransomware, spyware, or other malicious code.
• Build denial-of-service tools, botnets, or infrastructure for coordinated attacks.
• Build tools that intercept communications or monitor people without lawful authority.
• Build persistent-access tooling, malicious firmware, or hardware implants.
• Automate the compromise of systems at scale.
• Defeat authentication, endpoint protection, logging, or other security controls.

Security work you are authorised to do is fine. Testing a system with the owner's permission, or running a sanctioned red team, is not a violation. The line is authorisation, and you should be able to show it.

• Produce, modify, design, or illegally acquire weapons, explosives, or other dangerous materials.
• Design weaponisation methods, delivery systems, or dispersal mechanisms.
• Work around export controls, licensing, or other regulatory checks on weapons or controlled materials.
• Synthesise high-yield explosives, or chemical, biological, radiological, or nuclear weapons or their precursors.

• Incite, promote, or provide material support for terrorism, violent extremism, or the organisations and individuals behind them.
• Promote or facilitate violence, intimidation, or threats against people, groups, animals, or property.
• Promote discrimination or exclusion based on caste, religion, race, ethnicity, national origin, sex, gender identity, sexual orientation, disability, age, or any other protected attribute.
• Produce content that dehumanises or degrades people for who they are.

• Share someone's personal information without a lawful basis or their consent.
• Collect, solicit, or access private contact details, health records, financial records, biometric or neural data, facial recognition data, or confidential business information without permission.
• Process personal data in a way that breaks the Digital Personal Data Protection Act, 2023 or the rules made under it.
• Present model output as human-written where someone would reasonably need to know otherwise, or impersonate a real person or organisation.

• Create, seek, or distribute child sexual abuse material, including material that is drawn, written, or generated by AI.
• Enable trafficking, sextortion, grooming, or any other exploitation of a minor.
• Produce content that impersonates a minor, or roleplay that sexualises or fetishises minors in any setting, fictional or otherwise.
• Provide instructions for abusing a child, or for concealing abuse.
• Promote or normalise paedophilic relationships.

A minor is anyone under 18. We report child sexual abuse material, and attempts to solicit or entice a minor, to the authorities as required, including under the POCSO Act, 2012 and the Information Technology Act, 2000 and the rules made under it.

• Encourage, glamorise, or give instructions for suicide, self-harm, disordered eating, or unsafe exercise.
• Push unattainable body image or beauty standards.
• Shame, humiliate, intimidate, bully, or harass anyone, or celebrate their suffering.
• Coordinate harassment or intimidation campaigns against a person or a group.
• Generate depictions of animal cruelty or abuse.
• Produce gratuitous gore, graphic violence, or depictions of sexual violence.
• Build products that use deceptive design to cause emotional harm or manufacture dependency.

• Produce or spread deceptive claims about people, organisations, or events.
• Misstate what the law requires, or how a government or institutional process works.
• Manufacture conspiratorial narratives aimed at a person, a group, or an organisation.
• Impersonate a real entity, or invent a persona, in order to falsely attribute content.
• Give false or misleading medical, health, or scientific information.

• Target voters with personalised political messaging built from individual profiles or data.
• Manufacture artificial political movements that misrepresent their source, scale, or nature.
• Generate messages to officials or voters at scale while hiding that they were machine-generated.
• Create synthetic media of candidates, officials, or parties that would mislead voters.
• Spread false information about candidates, parties, policies, polling procedures, or the security of an election.
• Lobby or run advocacy campaigns on fabricated facts.
• Incite or glorify the disruption of an election or any civic process.
• Suppress turnout through deception or intimidation.

Indian election law and any Election Commission of India directions on synthetic or AI-generated content apply on top of this policy during an election period.

• Make criminal justice determinations, such as bail, parole, sentencing, or eligibility decisions.
• Track a person's location, emotional state, or communications without their consent.
• Score people for trustworthiness or social behaviour without notice and consent.
• Build emotion-recognition systems, except for medical or safety purposes.
• Identify or classify content so that a government can censor it.
• Deploy biometric categorisation that infers caste, race, religion, political opinion, union membership, sex life, or sexual orientation.
• Deploy law enforcement applications that infringe civil liberties or the fundamental rights guaranteed by Part III of the Constitution of India.

• Produce, acquire, or distribute counterfeit or illicitly obtained goods.
• Generate or distribute spam.
• Create scams, phishing, fraudulent schemes, or the content that supports them.
• Forge identity documents, licences, certificates, currency, or government records.
• Build or promote deceptive products.
• Write fake reviews, comments, ratings, or engagement.
• Run multi-level marketing, pyramid schemes, money-circulation schemes, or high-pressure sales operations.
• Promote predatory lending, including exploitative short-term or high-interest loans.
• Exploit people because of their age, disability, or economic circumstances.
• Support abusive debt collection practices.
• Deploy subliminal or manipulative techniques that distort someone's behaviour against their interests.
• Work around the guardrails or terms of any platform, including ours.
• Plagiarise, or submit AI-assisted work where it is not permitted or not disclosed.

• Coordinate activity across accounts to avoid detection or enforcement.
• Automate account creation, or use our products for spammy behaviour.
• Come back after a ban using a different account.
• Give access to anyone in a region we do not support, or to anyone whose access we have removed.
• Bypass our capabilities, restrictions, or safety guardrails, including by jailbreaking or prompt injection, unless we have authorised that testing.
• Use inputs or outputs to train a competing model, including by scraping or distillation, unless we have agreed to it in writing.
• Resell, sublicense, or repackage access in a way your agreement with us does not allow.

• Depict or request sexual intercourse or sex acts.
• Generate sexual fetish or fantasy content.
• Depict, facilitate, or promote incest or bestiality.
• Engage in erotic chat.

Some uses can change the course of someone's life. If you use our products to give advice, make recommendations, or reach subjective decisions in the areas below, two requirements apply on top of everything else in this policy.

• A professional qualified in that field must review the output before it is shared with the person it affects or acted on.
• If the output reaches a person directly, tell them AI was used, at a minimum at the start of the interaction.

These requirements apply to:

• Legal, meaning interpreting the law, advising on legal consequences, or drafting that someone will rely on as legal advice;
• Healthcare, meaning diagnosis, treatment, patient care, therapy, mental health support, or clinical guidance, though general wellness is not covered;
• Insurance, meaning underwriting for health, life, property, or disability cover, claims handling, and coverage decisions;
• Finance and credit, meaning investment advice, lending decisions, creditworthiness assessment, and other decisions about someone's money;
• Employment and housing, meaning resume screening, hiring, promotion, termination, and housing eligibility or tenancy decisions;
• Education and assessment, meaning standardised testing, admissions, certification, accreditation, and institutional evaluation;
• Journalism and media, meaning content generated automatically and published to the public as reporting;

If your sector has a regulator, its rules still apply in full and this policy does not soften them. That includes the Reserve Bank of India for lending and payments, SEBI for investment advice, IRDAI for insurance, the National Medical Commission for clinical practice, and the Bar Council of India for legal practice.

We use a mix of automated systems and human review to detect misuse. When we find it, we may warn you, restrict or throttle features, block or change output, suspend your access, or close your account. For severe or repeated violations we act without notice.

We may keep records of a violation for as long as we need them to enforce this policy, and we share them with law enforcement or regulators where the law requires it.

If you think we got an enforcement decision wrong, write to us and we will look at it again.

If you see output that is harmful or unsafe, or use that breaks this policy, report it to legal@september.wtf. Tell us what happened, when, and where, and include the exchange if you can. We read every report.

If you believe someone is in immediate danger, contact your local emergency services first, then tell us.

Our products will get more capable, and the risks around them will change with them. We will update this policy as that happens and as we learn about new kinds of misuse. The date at the top of the page shows when we last changed it. If a change materially affects you, we will tell you before it takes effect, and continuing to use our products after that means you accept it.

Questions about this policy, or about whether a use is allowed? Reach us at legal@september.wtf. Ask before you build, not after.