AI Output Disclaimer & Limitations

Last updated: 2026-08-01What our AI features can and cannot do, why output may be wrong or fabricated, that it is not professional advice, and the rules on how you may use what you generate.

AI Output Disclaimer & Limitations

Effective date: 1 August 2026 · Version 2.0

This notice explains the limitations of the AI features across the APERTURESyndicate platform, the rules on how you may use what they produce, and the legal terms that apply to AI output. Please read it before relying on anything an AI feature generates.

Where it applies. ASAI — assistant chat, roleplay, the Code agent, and image generation — ETHERANE, the AI tools inside ASVentor (background removal, object erase, outpainting, colourisation, restoration), and any other AI-assisted feature we ship. AI features are for users aged 16 and over.

Read this together with the Terms of Service (Section 7), AI Data & Training, Acceptable Use Policy, the ASAI Product Terms and, where financial topics are involved, the Financial Disclaimer.

We write these documents in good faith and keep them accurate to how the platform actually works. They have not yet been reviewed by qualified Estonian legal counsel.

1. Who provides these features

The provider is APERTURESyndicate OÜ (registry code 17384111, VAT EE102972654), Priisle tee 8, Lasnamäe linnaosa, Tallinn, Harju maakond, 13914, Estonia. Our single point of contact is [email protected].

We do not run the models. The computation that turns your input into a response happens at third-party inference providers acting as our sub-processors. What that means for your data — which category of provider receives what, in which country, and under which transfer safeguard — is set out in AI Data & Training and Data Processing & Sub-processors. We do not restate it here, so that it can never be out of date in two documents at once.

Which models sit behind ASAI Mini and ASAI Arc is not published: it is confidential to our business and it changes as the technology does. What matters for this document is that the output is generated, and the limits below apply whichever model produced it.

This has two consequences you should be aware of. Availability and behaviour of a model are partly outside our control, and a provider's own terms and safety filtering apply to your use (Section 9).

2. Output is provided as-is — no warranty of accuracy

AI features generate output by statistically predicting plausible text, code or images from your input. They do not know facts, do not understand truth, and do not verify what they produce. Output can be inaccurate, incomplete, outdated, biased, misleading or entirely fabricated — including confident-sounding citations, quotes, names, figures, code and images that are wholly invented. This is usually called "hallucination", and it happens most convincingly when the output looks authoritative.

To the maximum extent permitted by law, AI features and their output are provided "as is" and "as available", without warranty of accuracy, completeness, currency, reliability, fitness for a particular purpose or non-infringement. We do not warrant that any output is correct or suitable for your purpose, and we do not warrant uninterrupted availability.

This Section does not exclude or limit any liability that cannot be excluded under the mandatory law of Estonia or the European Union, and it does not affect the conformity rights you have as a consumer under Section 15 of the Terms of Service.

3. Output is not professional advice

AI output is not professional advice and must never be treated as such. It is not a substitute for legal, medical, psychological, financial, investment, tax, accounting, engineering, safety or any other professional or regulated advice, and using an AI feature creates no professional relationship.

Do not make consequential decisions on the basis of AI output. Consult a qualified professional and verify independently. Where output touches financial, investment or trading matters, the Financial Disclaimer also applies in full.

If you are in crisis, an AI feature is not the right tool. See Crisis & Safety Resources.

4. AI transparency

You are always interacting with an artificial-intelligence system, not a human being. AI features are labelled so this is never ambiguous, and we do not present model output as if a person wrote it.

Content produced by these features — text, code and AI-generated images — is synthetic. It may not reflect real people, real events, or reality at all. If you publish or distribute AI-generated content outside the platform, you are responsible for disclosing that it is artificial where the context, the law, or plain honesty toward your audience requires it. This obligation is becoming explicit under the EU AI Act for certain kinds of synthetic content, and it is good practice regardless.

5. Your responsibility for how you use output

You are responsible for how you use AI output and for the consequences. That means you:

  • read, review and independently verify output before relying on it, publishing it, or acting on it;
  • do not present output as fact, as human-authored, or as professional advice when it is none of those;
  • ensure your use complies with the law and with other people's rights; and
  • keep human judgement proportionate to the stakes — never hand a consequential decision to a model without meaningful review.

This applies with particular force to the ASAI Code agent: review, test and audit generated code before running it anywhere or shipping it. Generated code can contain errors, insecure patterns, or dependencies you did not intend. You remain responsible for what you execute.

6. Ownership and use of output

As between you and us, and subject to this notice and the Terms of Service, you may use the output you generate. We claim no ownership of it and we do not restrict your use of it.

Four limits are outside our control and you should understand them:

  • Your inputs. You must have the rights to everything you submit — prompts, files, code, reference images. You are responsible for making sure your inputs do not infringe anyone's intellectual property, privacy or other rights.
  • Copyright eligibility. In many jurisdictions, material generated by an AI system without sufficient human authorship may not attract copyright protection at all. We make no representation that any output is protectable or original, and identical or similar output may be generated for someone else.
  • No clearance. We do not warrant that output is free of third-party rights. Because output comes from a model trained on existing material, it can unintentionally resemble a protected work. Checking that your use is lawful is your responsibility.
  • Provider terms flow through (Section 9).

7. Prohibited uses

Your use of AI features is governed by the Acceptable Use Policy, which applies in full. Without limiting it, you must not use any AI feature to generate, request or distribute:

  • illegal content, or content that facilitates illegal activity;
  • non-consensual intimate imagery, or deceptive synthetic media of real, identifiable people;
  • content impersonating a real person or organisation, or presented so as to deceive about its source or authenticity;
  • disinformation, fraud, scams, or content intended to manipulate or defraud;
  • content infringing the intellectual property, privacy, publicity or other rights of any person;
  • malware, exploits, or instructions whose primary purpose is to cause harm; or
  • harassment, hateful content, or content inciting violence.

We do not pre-screen AI output. No filter of ours reads what a model returns before you see it — the model providers apply their own safety filtering, and that is a different thing from moderation by us. We act on reports about AI output exactly as we act on reports about any other content, and we may suspend, restrict or terminate access and remove content for breaches, and report unlawful activity to the authorities.

8. Image generation — additional rules

Image generation carries heightened risk and these rules are not negotiable:

  • No misuse of a real person's likeness. You may not generate images depicting a real, identifiable person without a lawful basis, and you may never generate intimate, sexualised, defamatory or deceptive imagery of a real person, including public figures.
  • No deceptive deepfakes. You may not create synthetic imagery designed to make people believe a real person did or said something they did not, or to misrepresent real events.
  • Zero tolerance for child sexual abuse material. Any attempt to generate, request or distribute CSAM or any sexualised depiction of a minor is absolutely prohibited. Such activity results in immediate termination, preservation of the material as the law requires, and a report to the competent authorities.
  • Disclosure. Treat and, where appropriate, disclose generated images as artificial (Section 4).

The same rules apply to the AI image tools in ASVentor, including erase, outpaint and restore — an edited photograph of a real person is still an image of a real person.

9. Third-party provider terms

Because inference runs at third-party providers, their acceptable-use and usage terms flow through to you and to your use of the output. You must not use our AI features in a way that would breach the applicable provider's terms. Where a provider's terms are more restrictive than ours, the more restrictive terms govern in practice.

Providers also apply their own safety filtering, which can refuse, alter or block a request or a response. A refusal is not necessarily a moderation decision by us.

10. Limitation of liability

To the maximum extent permitted by applicable law, we are not liable for loss or damage arising from AI output or from your use of or reliance on any AI feature — including decisions made or content published on the basis of output, inaccurate or fabricated or infringing output, code generated by the Code agent, or unavailability of AI features. Output is assistance; the risk of relying on it is yours.

Nothing here excludes or limits liability that cannot be excluded under the mandatory law of Estonia or the European Union — including liability for death or personal injury caused by negligence, for fraud, for intent or gross negligence, or under mandatory product-liability law. The full liability framework, including the cap and your consumer rights, is in Section 15 of the Terms of Service.

11. Changes

We may update this notice as our AI features, our providers, or our legal obligations change. Material changes are notified at least 30 days in advance under Section 17 of the Terms of Service, and the date at the top of this page always reflects the current version.

Contact

For any question about this notice, write to [email protected] in English or Estonian.

APERTURESyndicate OÜ Registry code 17384111 · VAT EE102972654 Priisle tee 8, Lasnamäe linnaosa, Tallinn, Harju maakond, 13914, Estonia Company registration details

This notice is published in English, which is the controlling language; any translation is for convenience only.

Version history

  • v2.0 — 2026-08-01 — Aligned with the rest of the AI documentation: product scope corrected (ETHERANE and the ASVentor AI tools added, the dead "ASAIStudio" name removed), the list of inference providers replaced by a pointer to Data Processing & Sub-processors so it cannot diverge, an explicit statement that we do not pre-screen AI output, image rules extended to the ASVentor editing tools, liability tied to the Terms of Service framework, and all company figures and contacts moved to shared placeholders.
  • v1.0 — 2026-06-22 — Initial publication.
AI Output Disclaimer & Limitations | AS Docs