Terms of Service

4.1 Prohibited Content

You will not, and will not permit your end users to, use the service to generate, solicit, distribute, or facilitate:

• (a) Child sexual abuse material (CSAM) or any content that sexualizes, exploits, or endangers minors in any form;
• (b) Non-consensual intimate imagery, sexual content depicting real individuals without verified consent, or deepfakes intended to deceive, defraud, harass, or impersonate;
• (c) Defamatory, harassing, threatening, or misleading content about identifiable real individuals, including political figures, including impersonation or fabrication of their statements;
• (d) Content that incites, glorifies, or facilitates violence, terrorism, genocide, or serious physical harm to people or animals;
• (e) Hate speech or content that demeans, dehumanizes, or targets individuals or groups based on protected characteristics including race, ethnicity, national origin, religion, disability, gender, gender identity, sexual orientation, or age;
• (f) Content designed to promote or facilitate self-harm, suicide, or eating disorders;
• (g) Sexually explicit content involving real people without consent, or any sexual content where participants are or appear to be minors.

4.2 Prohibited Activities

You will not, and will not permit your end users to, use the service to:

• (a) Provide medical, legal, mental-health, or financial advice directly to end users as a substitute for licensed professionals, or make consequential decisions about individuals (including in employment, credit, lending, insurance, housing, education, immigration, healthcare, or access to essential services) without qualified human review and appropriate disclosures;
• (b) Create, acquire, design, synthesize, or facilitate weapons, including chemical, biological, radiological, nuclear, or high-yield explosive (CBRNE) weapons, or provide operational guidance for attacks on critical infrastructure;
• (c) Develop, distribute, or deploy malicious code, malware, ransomware, spyware, or tools for unauthorized access to computer systems, networks, or data; conduct cyberattacks; or engage in fraud, phishing, social engineering, or identity theft;
• (d) Interfere with democratic processes, including generating or distributing disinformation about elections, candidates, voting procedures, or public officials; conduct coordinated inauthentic behavior or political influence operations;
• (e) Engage in unauthorized surveillance, tracking, profiling, biometric identification of individuals without consent, social scoring by governments, predictive policing based solely on profiling, or real-time remote biometric identification in publicly accessible spaces, except where strictly permitted by law;
• (f) Infringe intellectual property rights, publicity rights, or privacy rights of any third party, or process personal data in violation of applicable data protection laws (including GDPR, CCPA, Hong Kong PDPO, and equivalent regulations);
• (g) Violate applicable export control, sanctions, or trade laws, including by providing access to users in embargoed jurisdictions or to sanctioned individuals or entities;
• (h) Operate high-volume automated bots, spam campaigns, fake account generation, astroturfing, or coordinated inauthentic behavior on any platform;
• (i) Generate academic work, certifications, or credentials for fraudulent submission, or misrepresent AI-generated content as human-produced where disclosure is required by law or policy.

4.3 Platform Integrity

You will not, and will not permit your end users to:

• (a) Attempt to bypass, disable, circumvent, or manipulate any safety measures, content filters, moderation systems, rate limits, usage monitoring, or access controls implemented by AllToken or by the underlying model providers, including through prompt injection, jailbreak techniques, adversarial inputs, or fine-tuning aimed at producing prohibited outputs;
• (b) Reverse engineer, decompile, or extract the weights, training data, or proprietary components of any underlying model, or use outputs to train, develop, or improve competing AI models in violation of the relevant provider’s terms;
• (c) Resell, relicense, sublicense, or redistribute raw API access to underlying models in a manner that violates the terms of the respective model provider; AllToken is intended for use as an integrated routing and infrastructure platform, not as a bare API reseller;
• (d) Use the service to build applications, products, or services whose primary purpose falls within any prohibited category listed in this AUP or in any underlying provider’s policies;
• (e) Misrepresent AI-generated outputs in a way that could cause harm, including presenting AI output as authoritative human expert opinion in regulated domains.

4.4 Disclosure and Transparency Requirements

Where your application interacts with end users or the general public, you will:

• (a) Clearly disclose that users are interacting with an AI system when that is not otherwise obvious;
• (b) Not deceive users into believing they are interacting with a specific real person;
• (c) Provide appropriate disclaimers in domains that are regulated or safety-relevant;
• (d) Implement reasonable mechanisms for users to report abuse or harmful outputs generated through your application.

4.5 Your Obligations as an AllToken Customer

As a condition of your use of AllToken, you will:

• (a) Require your end users to agree to terms that pass through obligations substantively equivalent to this AUP, including all prohibited content and activities listed above;
• (b) Implement reasonable technical and organizational safeguards appropriate to your use case, including authentication, rate limiting, abuse monitoring, and content moderation where applicable;
• (c) Promptly investigate and act on reports of misuse by your end users, including suspending or terminating offending accounts;
• (d) Cooperate with AllToken and with underlying model providers in any investigation of suspected policy violations;
• (e) Not represent, directly or by implication, that your application or service is endorsed, certified, or affiliated with Anthropic, OpenAI, Google, AWS, Meta, Mistral, or any other underlying provider, unless expressly authorized in writing.

4.6 Enforcement

AllToken reserves the right to monitor usage for compliance with this AUP, investigate suspected violations, and take action including warning, rate-limiting, suspension, or termination of accounts that violate this policy or that create material legal, reputational, or operational risk to AllToken, its providers, its other customers, or third parties. We may also be required to share information about policy violations with underlying model providers or law enforcement as required by law or by provider agreements. We will not be liable for any loss resulting from enforcement actions taken in good faith under this section.

4.7 Image Generation Content Moderation Levels

When using AllToken’s image generation features (currently gpt-image-2 and any future image-generation models), you may select a moderation level (e.g., auto or low) in the Studio Advanced settings or via the moderation request parameter on the API.

AllToken passes your selected moderation level to the upstream image-generation provider (currently OpenAI) without modification. AllToken does not impose an additional layer of content filtering on top of the upstream provider’s enforcement.

By selecting a permissive moderation level such as low, you acknowledge and agree that:

• (a) The upstream provider’s permanent safety lines continue to apply at all moderation levels. These permanent safety lines include, but are not limited to: prohibition of child sexual abuse material (CSAM); prohibition of sexually explicit content depicting identifiable real persons without explicit consent and lawful authority; prohibition of content intended to incite real-world violence; and other categories enumerated in the upstream provider’s then-current Usage Policy (currently published at https://openai.com/policies/usage-policies/);
• (b) You are solely responsible for compliance with (i) the upstream provider’s Usage Policy, (ii) any applicable laws or regulations in your jurisdiction concerning the generation, storage, distribution, or display of the resulting content, and (iii) AllToken’s general Acceptable Use Policy in this Section 4;
• (c) AllToken may log, on a per-request basis, the moderation level you selected, your account identifier, the task identifier, and the timestamp of the request, for the limited purposes of abuse detection, billing reconciliation, and audit. AllToken does not log the prompt text or the generated image bytes for analytics or for any purpose beyond delivering the requested image to you.

4.8 Image-to-Image, Edit, and Variation Features

AllToken offers image-to-image features (including edit, mask edit, and variation modes) that accept user-uploaded source images and optional mask images as inputs and produce modified or derivative images as outputs.

By uploading any image to AllToken’s image-to-image features, you represent and warrant that:

• (a) Rights: You own all necessary rights to the uploaded image, or you have obtained a lawful license, permission, or other legal basis to upload it and to authorize AllToken and the upstream image-generation provider (currently OpenAI) to process it for the purpose of producing the requested edit or variation;
• (b) Real persons: If the uploaded image contains the likeness of a real, identifiable individual, you have either (i) the explicit and revocable consent of that individual to create the requested edit or variation, or (ii) other lawful authority to do so under your jurisdiction’s laws governing publicity, privacy, defamation, and synthetic media;
• (c) Lawful content: The uploaded image does not contain child sexual abuse material (CSAM), non-consensual intimate imagery, content depicting real-world violence intended to incite, or other content prohibited by the upstream provider’s then-current Usage Policy or by applicable law in your jurisdiction;
• (d) Resulting derivatives: You assume sole responsibility for any use, publication, redistribution, or display of the generated derivative image, including for compliance with copyright, trademark, publicity, privacy, defamation, AI-disclosure (where required by jurisdiction), and synthetic-media labeling obligations.

8.1 Model Provider Rights

The underlying AI models accessed through AllToken are the property of their respective providers and are governed by the providers’ own license terms. Your right to use outputs generated from a given model is subject to the terms of that model’s provider, in addition to these Terms. Some providers restrict commercial use, competitive use, or use of outputs for model training — you are responsible for reviewing and complying with the applicable provider terms for any model you use.