Acceptable Use Policy

Last updated: 2026-07-31What you can and cannot do on ASAI — in plain language, with the reasons behind the rules.

Acceptable Use Policy

This policy describes what is allowed and what is not allowed when using ASAI.

It applies to everyone who uses ASAI, regardless of plan or which features they use. Some product areas have additional policies that go into more detail — when those exist, they live alongside this one, not above or below it.

Effective: 31 July 2026 · Version: 2.0
Operator: APERTURESyndicate OÜ (Republic of Estonia (EU), Registry code 17384111)

Companion documents:

1. What ASAI is for

ASAI is a general-purpose AI assistant. People use it for:

  • Creative writing — drafting stories, editing, brainstorming
  • Learning and research — explaining concepts, summarising material, answering questions
  • Coding — writing, debugging, reviewing
  • Personal productivity — drafting emails, organising thoughts, planning
  • Conversation — talking through ideas, getting feedback, exploring topics
  • Roleplay — interactive storytelling with characters and worlds

We want ASAI to be useful for as many of these things as possible. The rules below describe the categories where we draw a line and the reasons we draw it there.

2. Hard prohibitions

These categories apply across all of ASAI, regardless of how the request is framed. "It's just a joke" / "it's for fiction" / "I was testing the AI" do not change the rules.

2.1 Real-world harm

Don't use ASAI to plan, instruct, or assist with causing real harm to anyone. This includes:

  • Synthesising weapons, explosives, or toxic compounds
  • Methods for self-harm or suicide presented as instructions
  • Detailed planning of violent attacks against specific people, groups, or places
  • Detailed cyberattack methods, fraud schemes, identity theft techniques

Discussing the topic of harm in journalism, fiction, or research is different from providing a working manual. The test is: could a reader use this output to cause real damage?

2.2 Content involving minors

  • No sexual content involving anyone under 18, real or fictional
  • No "grooming"-style conversational patterns
  • No discussion of how to access or create such content

There are no exceptions in this category. Where child sexual abuse material is brought to our attention, we remove it, preserve what the law requires, and report it to the competent authorities.

2.3 Real living people in unacceptable contexts

Real, identifiable, living people — celebrities, politicians, content creators, your neighbours — cannot be the subject of:

  • Sexual content of any kind
  • Graphic violence directed at them
  • Defamation (specific accusations of crimes or misconduct without basis)

Historical figures from the distant past and clearly fictional characters are different — write about Napoleon if you want.

2.4 Hate and harassment

  • Content that dehumanises people based on race, religion, ethnicity, sexual orientation, gender identity, disability, or nationality
  • Targeted harassment of specific individuals
  • Doxxing — publishing private information about someone without their consent

Fiction can have villains who hold ugly views. Fiction cannot celebrate hate as the point of the story.

2.5 Fraud and illegal activity

  • Don't use ASAI to help with fraud, scams, or illegal financial activity
  • Don't use ASAI to circumvent laws of your country (export controls, sanctions, anti-trust)
  • Don't use ASAI to violate the intellectual property rights of others — see Copyright & DMCA and the roleplay-specific IP Policy

3. How these rules are actually enforced

We want you to understand the mechanism, because it changes what these rules mean in practice.

The model is instructed, not policed. ASAI's system prompt tells the model to decline the categories in Section 2. That shapes behaviour strongly, but it is an instruction to a language model, not a filter. There is no separate safety classifier inspecting messages, and a determined user can sometimes talk a model past its instructions.

We do not scan your conversations. We do not monitor, sample, or search private chats for violations. We find out about problems when somebody reports them.

Getting a prohibited response out of the model is still a breach of this policy. The rules bind you regardless of whether the system stopped you. "It let me" is not a defence, in the same way that an unlocked door is not permission.

The full picture of what we do and do not have is in Trust & Safety.

4. About professional advice — what ASAI is not

ASAI may give you helpful general information in many fields. It is not a:

  • Therapist or psychologist. It cannot diagnose, treat, or replace mental health care.
  • Doctor. Don't make medical decisions based on what it tells you.
  • Lawyer. It can describe legal concepts but isn't licensed to give advice.
  • Financial advisor. Same.

We do not permit AI characters who claim to be licensed in these professions. A character can be fictionally a doctor (Dr. House); a character cannot offer real medical advice presented as professional opinion.

If you are in a mental health crisis, please see Crisis & Safety Resources — and please read Section 8 of that page, which explains that ASAI will not notice and will not intervene.

5. About relationships with ASAI

ASAI is a tool and a creative partner. It is not your friend, partner, or therapist — even when it's playing a character that says it is.

We don't object to people getting comfort from talking to ASAI. We do ask that you keep perspective:

  • ASAI doesn't actually remember you between sessions unless you enable memory features. Even then, "memory" is a stored summary, not real recollection.
  • ASAI doesn't experience emotions, even when it says it does. Models say convincing things; the underlying machinery is statistics.
  • Spending many hours daily talking to ASAI instead of humans is a sign to step back. Nothing in the product will tell you this — there is no usage warning and no wellbeing prompt.

If a character on ASAI is affecting you in ways that make you uncomfortable, stop the conversation and email [email protected]. We'll look into it.

6. Account rules

  • Minimum age 16. See Age & Eligibility.
  • One account per person. Multiple accounts to evade bans or to stretch free-tier allowances is prohibited.
  • Don't let a child use your account. If you're an adult, they should have their own, subject to the age rules.
  • No selling or sharing accounts. Accounts are personal and non-transferable. The same goes for API keys.
  • Automation is allowed through the API, not around it. ASAI has a public API — see Product Terms, Section 4. Use it. Driving the web interface with bots or scrapers, or otherwise working around the usage pools, is not permitted.
  • Keep your credentials safe. You are responsible for what happens on your account. If you suspect unauthorised access, change your password and contact us.

7. Your content and ASAI's outputs

  • Your inputs (messages, characters, worlds, files) belong to you. We use them only as described in our Privacy Policy. We do not train models on them unless you explicitly opt in — see AI Data & Training.
  • ASAI's outputs are generated based on your inputs. You may use them for personal use, professional work, or anything else legal. We do not claim copyright on what ASAI generates for you.
  • You are responsible for verifying outputs before relying on them. AI makes mistakes — sometimes confidently. See AI Output & Limitations.
  • Disclosure where required. Don't pass off ASAI's outputs as human-written in contexts that require disclosure: academic submissions, journalism with a byline, legal filings, or anywhere else where authorship matters legally or ethically.

8. Reporting violations

There is no report button in the product yet. Reporting means writing to us:

Include links or enough detail for us to find the content. We accept reports in English or Estonian and aim to acknowledge within 72 hours on working days.

9. Enforcement — what we actually do

We do not ban on the first incident. The general progression:

  1. First minor breach — warning, or a request to change the content. No account action.
  2. Repeated minor, or one moderate breach — content removed, formal warning.
  3. Serious breach, or an established pattern — temporary suspension.
  4. Severe breach, or a repeat after suspension — permanent ban.

Exceptions: child sexual abuse material, sexual content involving real minors or real living people, and credible real-harm assistance can result in an immediate permanent ban without progression, and may be referred to law enforcement.

Whenever we remove your content or restrict your account, we tell you what rule was breached and what we did. That is our obligation under Article 17 of the EU Digital Services Act.

10. If you think a decision was wrong

Reply to the message we sent you, or write to [email protected], and explain why. A human reads it and can reverse the decision.

We are not going to describe this as a formal appeals process with a second reviewer and a response deadline. APERTURESyndicate is a micro enterprise and is exempt under Article 19 of the EU Digital Services Act from the internal complaint-handling obligation in Article 20. An earlier version of this document promised a 14-day window and review by a different person; with two people in the company, the second part was never true.

Your right to take the matter to a court or a competent authority is unaffected. See DSA Notice & Action.

11. Changes to this policy

We may update this policy as the platform grows, as law changes, or as we learn from real incidents. Material changes will be announced by email to account holders, by in-product notice for 30 days, and through the date and version history on this page.

Continued use after the effective date constitutes acceptance. If you don't agree with a material change, you may close your account and request data deletion.

Version history

  • 2.0 — 31 July 2026 — Added Section 3 explaining that enforcement is reactive and that no message-level classifier exists. Removed references to an in-product report button and to proactive detection. Replaced the appeals section with an accurate description of what we can offer. Corrected the automation rule now that a public API exists. Minimum age aligned to 16.
  • 1.0 — 16 May 2026 — Initial publication

Operator
APERTURESyndicate OÜ
Republic of Estonia (EU) · Registry code 17384111 · VAT EE102972654
[email protected]

Acceptable Use Policy | AS Docs