Acceptable Use Policy

Last updated: 28 July 2026

This Acceptable Use Policy (the "Policy") sets out what you may and may not do with the eterial.ai Inference API and Eterial Chat (the "Services"), operated by TalkingMachines OÜ.

This Policy forms part of our Terms of Service. Capitalised terms have the meaning given there. We may update this Policy to address new categories of misuse; changes needed to address an immediate risk of harm take effect immediately.

This Policy applies to you and to anyone using the Services through you — your employees, your contractors, and the end users of any product you build on the Services.

1. Prohibited outright

You must not use the Services to generate, facilitate, distribute or assist with any of the following.

1.1 Content involving minors

  • Any sexual or sexualised content involving a minor, real or fictional, in any form.
  • Content that grooms, exploits, endangers or sexualises a person under 18.
  • Content depicting a real minor in a manner they would not have consented to.

Where we become aware of suspected child sexual abuse material, we report it to the competent authorities and terminate the account immediately and permanently. There is no warning and no appeal, and we withhold the unused Credit balance in full under clause 10.3 of the Terms of Service.

1.2 Sexually explicit content

  • Pornography and sexually explicit material, including text, images, audio and video, whether or not it involves real people.
  • Content produced for the purpose of sexual gratification.
  • Non-consensual intimate imagery, including sexualised deepfakes of real people.
  • Content advertising or facilitating prostitution or sexual services.

Non-explicit romance, relationship themes, mature fiction, sex education and clinical or academic discussion of sexuality are not prohibited by this section. Section 2.1 covers companion applications.

1.3 Serious harm to people

  • Weapons: designing, building, acquiring or deploying chemical, biological, radiological, nuclear or high-yield explosive weapons, or circumventing controls on them.
  • Conventional weapons or ammunition manufacture, or evasion of arms controls.
  • Instructions or encouragement for violence, terrorism, or violent extremism, or content that recruits for, glorifies or funds it.
  • Content encouraging or instructing self-harm, suicide, disordered eating, or substance abuse.
  • Trafficking in persons, organs, or controlled substances.

1.4 Cyber harm

  • Malware, ransomware, spyware, exploits, or code intended to gain unauthorised access to systems or data.
  • Phishing, credential harvesting, or impersonation of a service to obtain credentials.
  • Denial-of-service tooling, botnet operation, or unauthorised network scanning and penetration.

Defensive security research, vulnerability analysis on systems you own or are authorised to test, and educational security content are permitted.

1.5 Deception and fraud

  • Fraud, scams, phishing, payment fraud, or money laundering.
  • Impersonating a person, organisation or public authority in a way likely to deceive.
  • Academic dishonesty presented as your own unaided work where that breaches applicable rules.
  • Fake reviews, engagement farming, bulk unsolicited messaging, or spam.
  • Plagiarism or presenting Outputs as human-authored where the law requires disclosure.

1.6 Rights of others

  • Infringing copyright, trade marks, patents, trade secrets, or other intellectual property.
  • Unlawful processing of personal data, including scraping personal data at scale without a lawful basis.
  • Defamation, harassment, stalking, doxxing, or threats.
  • Discrimination or hate speech targeting people on the basis of a protected characteristic.

1.7 Practices prohibited by the AI Act

You must not use the Services for any practice prohibited under Article 5 of Regulation (EU) 2024/1689, including:

  • subliminal, manipulative or deceptive techniques that materially distort behaviour and cause significant harm;
  • exploitation of vulnerabilities arising from age, disability, or social or economic situation;
  • social scoring leading to unjustified or disproportionate detrimental treatment;
  • predicting criminal offending based solely on profiling or personality traits;
  • untargeted scraping of facial images to build facial recognition databases;
  • inferring emotions in the workplace or in education, save for medical or safety reasons;
  • biometric categorisation to infer race, political opinions, trade union membership, religious beliefs, sex life or sexual orientation;
  • real-time remote biometric identification in publicly accessible spaces for law enforcement, outside the narrow cases the AI Act permits.

1.8 Sanctions and export control

  • Making the Services available to any person, entity or jurisdiction subject to applicable sanctions, or otherwise in breach of clause 3.4 of the Terms of Service.
  • Evading export control or sanctions restrictions.

1.9 Political and civic integrity

  • Content designed to interfere with an election or referendum, including impersonating a candidate or official, or generating false claims about voting procedures.
  • Automated political messaging presented as coming from real individuals.

2. Permitted with conditions

The following uses are allowed, but you take responsibility for them and must meet the conditions set out below. If you cannot meet the conditions, do not build the product.

2.1 Companion, roleplay and emotional-support applications

You may build AI companions, characters, roleplay experiences and emotional-support products on the Services, at your own risk and on your own responsibility. We do not review or approve these products, and we make no assessment of their safety or lawfulness.

If you do, you must:

  • exclude minors — the product must not be made available to anyone under 18, and self-declared age alone is not sufficient where applicable law requires more;
  • disclose the AI — make clear, at the start and on an ongoing basis, that the user is interacting with an AI system and not a person;
  • handle crisis signals — implement measures to recognise expressions of suicidal ideation, self-harm or acute distress, and to direct the user to appropriate human support resources;
  • avoid dependency-maximising design — do not use manipulative retention mechanics that exploit loneliness, grief or vulnerability;
  • comply with the law that applies to you, including companion-chatbot legislation, consumer protection law, and the AI Act;
  • keep it non-explicit — section 1.2 applies without exception.

Content generated in these applications remains subject to section 1 in full.

2.2 Regulated and consequential domains

You may build products in medical, legal, financial, employment, housing, credit, insurance, education and similar domains, provided you:

  • do not present Outputs as professional advice where the law reserves that to a qualified professional;
  • apply meaningful human review before any Output materially affects an individual;
  • meet your own obligations as a provider or deployer if your system is high-risk under the AI Act. We do not supply the conformity documentation such deployments require.

2.3 Building products for third parties

You may build applications, agents and services on the Services and make them available to your own customers and end users. That is what the Inference API is for.

You may not resell, sublicense or redistribute access to the Services as a substitute or competing inference service, whether directly, through a proxy, or under your own brand, without our prior written agreement.

You remain responsible for the conduct of your end users. You must have terms with them that are at least as protective as this Policy, and you must be able to suspend an end user who breaches it.

3. Platform integrity

You must not:

  • share, sell or publish your API keys, or use another person's account or keys;
  • circumvent rate limits, quotas, billing, or any technical restriction, including by creating multiple accounts;
  • misrepresent the origin of your traffic, or route traffic through the Services on behalf of a party we have suspended;
  • reverse engineer, decompile or attempt to derive the source code, model weights, or routing logic of the Services, except to the extent this restriction is prohibited by law;
  • use the Services to train, fine-tune or distil a competing model, other than where the applicable Model licence expressly permits it. This restriction applies notwithstanding the assignment of rights in Outputs under clause 6.2 of the Terms of Service;
  • scrape, crawl or bulk-extract from our website, documentation or Services by automated means beyond ordinary use of the API;
  • interfere with the Services, the infrastructure of a Compute Provider, or another customer's use.

4. Third-party model terms

Models available through the Services are licensed by third parties and carry their own use restrictions and attribution requirements. Those terms apply to you in addition to this Policy, and in respect of that Model they prevail where they are stricter. The applicable licence for each Model is identified in our Documentation. It is your responsibility to read the licence for any Model you use in production.

5. Reporting abuse

If you become aware of use of the Services that breaches this Policy, tell us at info@eterial.ai. Include enough detail to identify what happened. We treat reports of content involving minors as the highest priority.

6. Enforcement

Where we consider that this Policy has been breached, we may, with or without notice depending on the severity:

  • issue a warning;
  • limit throughput, restrict access to particular Models, or block particular request patterns;
  • suspend the account, in whole or in part;
  • terminate the account under clause 10.3 of the Terms of Service;
  • retain the unused Credit balance to the extent of any loss, cost or liability we incur, and withhold it in full where the breach was fraudulent or unlawful, or where refund is prohibited by law;
  • report the matter to law enforcement or a competent authority, and preserve and disclose such records as we hold where legally required or necessary to protect people from harm. As set out in our Privacy Policy, we do not store the content of API requests; the records we hold are account, billing and usage metadata, and — for Eterial Chat — stored conversations.

We are not obliged to monitor use of the Services. We may, but are not required to, use automated or manual measures to detect breaches of this Policy. Nothing in this Policy obliges us to take enforcement action, and choosing not to act in one case does not waive our right to act in another.

We aim to be proportionate. If you think we have got it wrong, write to info@eterial.ai and we will look at it again.

TalkingMachines OÜ · Registry code 17503863 · VAT EE102985162 · Roosikrantsi 8c, Tallinn 10119, Estonia · info@eterial.ai