Acceptable Use Policy

This Acceptable Use Policy ("AUP") is part of the Audexum Agreement and is binding on every user of the Service. It explains what you may not do with Audexum. Violations are a material breach of the Terms of Service and may result in suspension, termination, forfeiture of fees, reporting to law enforcement, and civil claims. We may update this AUP at any time; the current version is the one posted at audexum.com/aup.

The default text-to-speech model powering the Service ("Supertonic-3") is distributed under the BigScience Open RAIL-M licence. That licence (Attachment A) requires us to pass certain use restrictions downstream to every person who uses the model's output. By using the Service you agree to all of the following restrictions, which flow directly from the model licence and which Audexum is contractually required to enforce:

You must not use the Service, or any Output generated by the Service, for any of the following purposes:

1. Any use that violates applicable national, regional, or international laws or regulations, including data-protection, privacy, and export-control law.
2. Any use intended to, or reasonably likely to, cause harm to, or exploit, minors in any way, including generating or distributing content that sexualises or endangers children.
3. Generating or distributing verifiably false information and/or content that is presented as factual, with the purpose of deceiving, radicalising, or spreading propaganda.
4. Generating or distributing personal information with the purpose of harassing, targeting, or threatening specific individuals.
5. Providing automated decisional output that produces legal or similarly significant effects on individuals — such as denying employment, credit, housing, insurance, or public services — without adequate human oversight, review, and the ability to appeal or contest the decision.
6. Discriminating against or exploiting individuals or groups based on legally protected characteristics, including age, disability, ethnicity, gender, gender identity, national origin, race, religion, sex, or sexual orientation.
7. Providing, or holding out as providing, medical, legal, financial, or other professional advice that requires licensed expertise, in contexts where reliance on automated output could cause real harm to individuals.
8. Intentionally targeting and/or exploiting vulnerable groups — including the elderly, children, people in mental-health crises, or people in financial distress — in ways that increase their vulnerability or cause harm.
9. Any use whose primary purpose or likely effect is to create unnecessary harm, threat, surveillance, or illegitimate distress to any person or group.

These restrictions apply in addition to all other sections of this AUP and the Terms of Service. They cannot be waived by agreement, plan tier, or enterprise contract. Violation of any item in the list above is a material breach of the Agreement and grounds for immediate termination of your Account, forfeiture of unused balance, and potential referral to competent authorities.

Other models available through the Service (including Kokoro voices) are distributed under separate licences (Apache 2.0 and others); their use is governed by this AUP and the applicable licence terms. Where multiple models are used in a single request, the most restrictive applicable terms govern.

You must not use the Service to:

• generate audio that purports to be, or could reasonably be mistaken for, a real natural person — living or deceased — without that person's (or their estate's) prior, explicit, informed, written consent;
• generate audio depicting any politician, head of state, election candidate, public official, judge, regulator, or any person discharging a public function, in any manner reasonably capable of misleading a listener as to authenticity, regardless of consent;
• generate "deepfake" audio, voice clones, or imitations of real persons; the Service does not currently offer voice cloning, and you must not use the preset voices to mimic, parody as if real, or otherwise impersonate a specific real person;
• misrepresent the source of the audio (for example, presenting AI-generated audio as a human recording where listeners would care about the distinction);
• impersonate Audexum, our personnel, or any other person or organisation;
• fail to disclose to listeners, where required by Article 50 of the EU AI Act or any other applicable law, that the audio is AI-generated.

The single highest-risk use cases for Audexum — election interference, financial fraud through voice impersonation ("CEO voice scam"), and non-consensual intimate audio — are categorically prohibited and will be reported to law enforcement.

You must clearly and prominently disclose to your end users — before or at the point they first hear the audio — that the audio was generated by artificial intelligence. This obligation applies regardless of whether your jurisdiction currently mandates such disclosure by law. Acceptable forms of disclosure include a visible label, an accompanying text notice, or an audible announcement at the start of the content.

You may not present Output as a recording of a live or pre-recorded human voice in any context where a reasonable listener would consider the distinction material — including but not limited to customer-service calls, medical advice, legal advice, educational content directed at minors, news broadcasts, and financial guidance.

By using the Service, you accept sole responsibility for ensuring your end users are aware they are listening to AI-generated audio. Audexum accepts no liability for any harm arising from your failure to make this disclosure.

You must not generate, distribute, or use the Service to facilitate the generation of audio that sexualises any person who is, or appears to be, under the age of 18. This includes content depicting child sexual abuse material ("CSAM") in audio form, sexual roleplay involving minors, or grooming scripts targeting minors. We permanently terminate offending accounts, retain associated logs for the maximum time permitted by law, report to law enforcement (including the Bulgarian Cybercrime Unit and INHOPE / IWF where appropriate), and cooperate fully with any investigation. No refund, no appeal.

You must not use the Service for, or to facilitate:

• phishing, vishing, smishing, or any social-engineering scheme;
• "grandparent" scams, fake-kidnapping scams, or other voice-based scams targeting individuals;
• impersonation of banks, payment processors, government agencies, tax authorities, customs, police, courier services, or utility companies in communications with their customers;
• generation of fake testimonials, fake reviews, fake interviews, fake news anchors, or fake newsroom audio;
• generation of audio used to manipulate financial markets, perform pump-and-dump schemes, or commit insider trading;
• unsolicited robocalling, automated calling, or interactive-voice-response ("IVR") campaigns that violate the EU ePrivacy Directive, GDPR, the US TCPA, the UK PECR, or any equivalent telecommunications law;
• collection of personal data through deceptive audio prompts.

You must not generate or distribute audio that:

• incites violence, terrorism, genocide, ethnic cleansing, or material harm to any person or group;
• promotes self-harm, suicide, or eating disorders;
• constitutes harassment, threats, stalking, doxxing, or non-consensual disclosure of personal information;
• constitutes hate speech directed at protected characteristics including race, ethnicity, national origin, religion, disability, gender, gender identity, or sexual orientation;
• infringes any third party's intellectual property, trademark, publicity, or privacy right (including reading from copyrighted works you do not have rights to);
• contains malware, malicious code, or executable payloads in any form (we still police the Input even though Output is audio);
• is otherwise unlawful in any jurisdiction in which the Output will be distributed.

You must not:

• create multiple Accounts to circumvent free-tier limits, payment failures, or suspensions;
• share, sell, lease, or expose your API Key to end users or any third party that is not your employee or authorised contractor under equivalent confidentiality obligations;
• scrape, crawl, mirror, automate against, fuzz, brute-force, or otherwise attempt to extract bulk data, voice embeddings, model weights, or training data from the Service;
• reverse-engineer, decompile, disassemble, or attempt to derive the source code of the Service or underlying models, except to the limited extent expressly permitted by mandatory law that may not be contracted out of;
• use the Service, the Input, the Output, or any related data to train, fine-tune, evaluate, distil, or otherwise develop a competing text-to-speech, speech synthesis, voice cloning, voice conversion, or generative audio model or service;
• probe, scan, or test the vulnerability of any of our systems or networks without prior written authorisation from [email protected];
• interfere with or disrupt the Service, including by mounting denial-of-service attacks, distributing malware, or sending traffic at rates clearly exceeding what your plan permits;
• attempt to bypass quotas, rate limits, paywalls, authentication, content filters, or any access-control mechanism;
• frame, mirror, or republish the Service or any substantial portion of it as if it were your own product;
• remove, alter, conceal, or disable any provenance marker, watermark, metadata tag, or notice that we embed in the Output.

If you deploy the Output through telephone, SMS, paging, public-address systems, radio, television, or streaming services, you are solely responsible for compliance with all telecommunications and broadcast laws, including the EU ePrivacy Directive (and the forthcoming ePrivacy Regulation), Bulgarian Electronic Communications Act, the US Telephone Consumer Protection Act ("TCPA"), the UK Privacy and Electronic Communications Regulations ("PECR"), Canada's CRTC Unsolicited Telecommunications Rules, and Australia's Spam Act. We do not warrant Output is suitable for any such channel. You must not present Output as live human speech where a regulator, contract counterparty, or audience would care.

You must not submit Input that:

• contains personal data of third parties without a lawful basis under Article 6 GDPR (and, for special categories, Article 9);
• contains non-public personally identifiable information (financial-account numbers, payment card data, government identifiers, biometric data, health data, sexual orientation, precise geolocation) of any individual without explicit consent;
• has been obtained through unlawful means (data breaches, scraped data covered by anti-scraping injunctions, stolen credentials).

You must not direct the Service to children under 16. You must not generate audio intended to deceive, manipulate, or sexually solicit minors. If the Service is used in any educational product, you must obtain appropriate consents under your local children's-privacy law (including KOPPA / COPPA where applicable).

The Service is not designed or warranted for use in:

• emergency communications (including 112 / 999 / 911 dispatch);
• life-support, surgical, or diagnostic medical systems;
• autonomous-vehicle command, aviation control, or maritime navigation;
• nuclear-facility control;
• critical-infrastructure control systems;
• any safety-critical system where failure or audio inaccuracy could cause death, serious injury, environmental damage, or critical infrastructure failure.

You use the Service in such applications at your own risk and with no warranty, express or implied, from us.

You must not use the Service, the Output, or any related technology:

• in or for the benefit of any country, person, or entity subject to comprehensive EU, UK, US, or UN sanctions;
• in any manner that would violate EU dual-use export controls under Regulation (EU) 2021/821 or the US Export Administration Regulations.

To report suspected misuse of the Service, including impersonation, deepfakes, fraud, harassment, or CSAM, email [email protected] with as much detail as you can share. We aim to acknowledge reports within 48 hours and take appropriate action, which may include account suspension, content takedown, and referral to law enforcement.

Trusted flaggers and rights-holders may also use [email protected] to submit DMCA-style notices and intellectual-property infringement claims, including the standard elements: identification of the work, the allegedly infringing use, your contact details, a statement of good-faith belief, and a signed statement under penalty of perjury.

Violations may be enforced at our sole discretion, with or without notice, and may include any combination of:

• warnings;
• temporary rate-limit reductions;
• suspension of API Keys, Studio access, or the Account;
• permanent termination of the Account;
• blocking your payment method and IP ranges;
• forfeiture of prepaid fees;
• preservation and disclosure of logs to law enforcement;
• civil claims for damages and injunctive relief;
• criminal referral.

We may also reverse-charge any fees, fines, or costs we incur as a result of your violation, including charge-backs and provider penalties from Stripe, Cloudflare, or other sub-processors.

We will cooperate with valid legal process from competent authorities, including subpoenas, court orders, and law-enforcement requests, and we may notify you of such requests where we are legally permitted to do so. For CSAM and credible threats of imminent harm, we may disclose information without prior notice.

We may update this AUP at any time to address new threats, regulatory developments, or product changes. Material changes are announced on the site at least 30 days in advance. Continued use after the effective date constitutes acceptance.

• Report abuse: [email protected]
• Legal / DMCA: [email protected]
• Security: [email protected]