Acceptable Use Policy
Last updated September 15, 2026
The rules that calls made through Amux must follow, and what happens when they are broken.
This policy is issued by Amux AI, forms part of the Terms of Service, and applies to all calls made through Amux. Breaching it may result in your service being restricted or terminated.
Scope
Amux is a conduit, not a destination. Your requests are forwarded to upstream model providers, so two sets of obligations apply to you at once:
- This policy and the Terms of Service
- The usage policy of the upstream provider you call
The second does not lapse because the call went through Amux. Policies for the same model can differ between providers, so check before pinning one.
You are responsible for all content sent through your account and keys, including calls made by others using a key you leaked.
⚠️ The service is not offered in mainland China. For the geographic restriction, what counts as circumvention, and how it is enforced, see Section 15 of the Terms of Service.
Prohibited uses
Unlawful content
- Content that violates any applicable law or regulation
- Sexualized content involving minors — zero tolerance; accounts are banned immediately and reported as required by law
- Non-consensual intimate imagery, or content used to harass or humiliate a specific individual
- Terrorist or extremist propaganda, or content inciting violence or hatred
Infringing the rights of others
- Content that infringes intellectual property, trade secrets, or privacy rights
- Generating identifiable likenesses of real people without consent, or impersonating individuals, organizations, or government bodies
- Processing others' personal information without a lawful basis
Deception and manipulation
- Generating fake news, forged records, forged credentials, or fake reviews
- Phishing content, fraud scripts, or content designed to defeat identity verification and risk controls
- Generating and distributing content at scale to manipulate public opinion or search results
- Concealing that content is AI-generated when interacting with real people, where applicable law requires disclosure
Security and abuse
- Developing or distributing malware, ransomware, or exploit tooling
- Unauthorized intrusion, scanning, crawling, or stress testing
- Bypassing spend limits, forging call origins, or otherwise interfering with the operation of the service
- Circumventing our content safety mechanisms or those of upstream providers
- Circumventing the geographic restriction, including by VPN, proxy, false identity, or nominee accounts — the service is not offered in mainland China; see Section 15 of the Terms of Service
High-risk applications
Without holding the relevant qualifications and accepting the corresponding responsibility, model output must not be used directly for:
- Medical diagnosis, treatment recommendations, or medication guidance
- Legal advice
- Financial investment advice, credit, or insurance decisions
- Automated decisions with a significant effect on individuals, such as employment, education admissions, or housing
- Control of critical infrastructure or safety-related systems
- Weapons development, or any purpose related to weapons of mass destruction
These applications are not categorically prohibited, but a qualified person must review the output; model output may not be the final decision.
Commercial restrictions
- The service may not be resold, or provided to third parties in a manner substantially equivalent to resale, without written consent
- The service may not be used to train competing models
Generated content
This policy applies to generated content in every modality — text, images, video, audio, and anything derived from them. The prohibited uses above apply equally to images and video, and several of them carry heavier consequences in visual form.
We do not review content in advance
We do not review the requests you send, nor the content models return. As a calling conduit, the platform performs no substantive review of content before it reaches the upstream provider. That means:
- Whether content is lawful depends on your own judgment and the law that applies to you
- Responsibility for violating content rests with the account that made the call
- We investigate on receiving a report, on detecting anomalies, or when required by a competent authority
Some models and providers operate their own safety mechanisms and may refuse to generate certain content. Such refusals come from upstream, not from us; the platform returns the upstream refusal reason as given, with no silent substitution. Policies for the same model can differ between providers.
Particular care with images and video
Misuse of visual content usually has heavier consequences than text. Take particular care with the following:
| Type | Requirement |
|---|---|
| Identifiable real people | Do not generate a person's likeness without their consent, and in particular do not generate imagery that could be mistaken for something that actually happened |
| Public figures | Do not generate content that could be mistaken for their real words or actions; satirical and commentary work should be clearly labelled |
| Adult content | Upstream policies differ and most prohibit it outright; anything involving minors is zero tolerance |
| Protected characters and trademarks | Generating well-known characters or brand marks may infringe third-party rights; obtain a licence before commercial use |
| Documents and credentials | Do not generate identity documents, receipts, seals, or other artifacts usable for deception |
Where applicable law requires AI-generated content to be labelled, label it. Some models offer a watermark parameter, but the presence or absence of a watermark does not change your labelling obligation.
Generated artifacts live in publicly readable storage
The address of an image or video artifact is unguessable, but anyone who obtains it can access it without signing in. Once you have generated something that should not be seen by others, a leaked address cannot be recalled.
Do not treat "nobody knows the address" as a protection. Where access control matters, retrieve the artifact and store it yourself. See data retention and privacy controls.
Rights in output, and the risks
Model output is produced by a probabilistic model: it may be identical to another user's output, may not be eligible for copyright protection, and may resemble existing works, trademarks, or a person's likeness. Check before using it externally; see Section 9 of the Terms of Service.
Your responsibilities
When using the service, you should:
- Protect your keys. A key is the calling identity, and charges arising from a leak are the account holder's responsibility
- Take responsibility for output. Models can produce inaccurate, biased, or inappropriate content; review it before putting it in front of others. This applies to images and video as much as to text
- Take responsibility for where artifacts end up. Generated artifacts live in publicly readable storage, so do not leave anything there that should not travel
- Disclose where required. Where applicable law requires you to say that content is AI-generated, say so
- Comply with upstream policies. See "Scope" above
Reporting
To report abuse, open a ticket under the "report abuse" category, or email support@amux.ai. Reports can be raised from any account type.
Please include specifics — the account involved, times, content, or call records — so that we can investigate.
Enforcement
Depending on severity, we may take the following measures:
| Measure | When |
|---|---|
| Notice and request to remediate | Minor or apparently unintentional breaches |
| Restricting specific models or capabilities | Repeated problems of the same kind |
| Suspending the account | Serious breaches, or as an interim measure during investigation |
| Terminating and banning | Severe or repeated breaches, zero-tolerance content, or circumventing the geographic restriction |
Where service is terminated for a breach of this policy, the related fees are not refundable (see the Terms of Service).
Where required by law or by a competent authority, we will cooperate with investigations and provide the necessary information.
Appeals
If your account is suspended or banned, you may appeal through a ticket or by emailing support@amux.ai with an explanation. We will review and respond.
Changes to this policy
When this policy changes, we update the effective date at the top of this page. Where the change is material, we give notice by email or in the console.