Spiral is a messenger with a personal AI assistant, built on the Matrix protocol. The Spiral service is provided by Andrii Panasiuk, Ukraine ("we", "us" below). Contact: Spiral Support, support@spiral.md.
These terms apply to the Spiral iOS app and the Spiral service at spiral.md (together, "Spiral"). Our Privacy Policy explains what happens to your data and is part of these terms. If you downloaded the app from the App Store, Apple's Licensed Application End User License Agreement (the Standard EULA) also applies.
You are asked to accept these terms when you create an account, and by using Spiral you accept them. If you do not accept them, do not use Spiral.
2. Who can use Spiral
You must be at least 16 years old.
Registration is by invitation: you need an invitation link from an existing user or from us.
You may not use Spiral if we have banned you before, or where the law forbids you to.
3. Your account
Your account is created and signed into with a recovery phrase made on your device, and optionally a passkey. There is no password, email address or phone number.
Anyone who has your recovery phrase can sign in as you. Keep it secret. We never ask for it and cannot see it.
You can get a new phrase from a device where you are signed in; a reset link also needs your account's passkey.
You are responsible for what happens under your account.
4. What Spiral does
Chats and calls. Messages and calls between people's devices are end-to-end encrypted.
The feed. The Feed tab is one public stream inside the app: every Spiral account can read the posts and comments in it, and nobody without an account can. In this version posts are text only (up to 5,000 characters; comments up to 1,000), shown newest first. You can like a post, comment on it, share it into a Spiral chat, hide it, and report or block its author. Who liked a post is shown to everyone who can see the post. The feed is not end-to-end encrypted: we store what is in it readable, so that everyone can see it and we can moderate it (section 6). Before your first post or comment you accept the feed rules, a short summary of the rules of section 6.
Your assistant. Every account has a personal AI assistant. It is a separate account with its own name: what it writes is shown as written by your assistant, not by you. It answers you in your chat with it; in a group it answers when someone mentions it, and it runs the room tools you set up there (for example translating messages, recording or summarising calls). When someone writes to it for the first time, it asks you before it answers. Assistants do not read or post in the feed.
Other people's assistants can be members of chats you are in. They act for their owners.
Assistants in calls. An assistant in a call either talks with its owner, or stays silent while recording, summarising or interpreting. Every participant can see which assistants are in the call and whether one is recording.
Interactive posts. Your assistant can write small interactive posts (polls, games, cards) on request. You decide whether to publish one. A published post is a message of the chat it is posted in, and it runs on the reader's device in a sandbox without network access. Interactive posts cannot be published in the feed.
5. Third-party AI and your consent
The assistants' AI runs on services of OpenAI and Google. Your messages and voice are sent to them only after you allow it in the app, and you can withdraw at any time in Settings → Privacy → Third-party AI. A withdrawal during a call ends the AI in that call at once.
Without your consent your own assistant is silent, no assistant acts on your messages, and your messages are left out of what is sent to an AI. A call has no AI unless everyone in it agreed. Chats and calls keep working.
Nothing in the feed is sent to an AI, and no AI decides about it.
AI can be wrong, incomplete or inappropriate. Check what an assistant writes before you rely on it, and do not use it for medical, legal, financial or emergency decisions.
Each assistant has a weekly AI limit. The week starts on Monday at 00:00 UTC; the app shows when the limit renews in your local time. Within the weekly limit we may also limit a kind of use (for example calls or call translation), and we may change the limits (section 11). The app shows how much of a limit is used as a share, not as an amount of money. How fast a limit is used depends on what the AI costs to run, so a limit is not a promise of a number of messages or minutes. When the weekly limit or the limit on a kind of use is reached, the assistant stops that use at once — in a call too, while the call itself goes on — and the app tells you. It works again when the limit renews or when we lift it. The limit does not apply to chats and calls between people.
6. Rules for content and behaviour — zero tolerance
Spiral has zero tolerance for objectionable content and abusive users. You must not use Spiral — or make an assistant, a post, a comment or a call carry — anything that breaks one of these rules. Each rule has the name our notices use when we act on it.
Illegal content — anything that is illegal, or encourages or helps someone break the law, including offering illegal goods or services, and promoting, facilitating or profiting from human trafficking, forced labour or other exploitation of people. Under Ukraine's martial law this includes publishing the positions or movements of armed forces or the places of strikes, and justifying or glorifying the armed aggression against Ukraine.
Regulated goods — selling or offering drugs, weapons or other regulated goods.
Child safety — anything that sexually exploits or endangers children, in any form.
Sexual content — pornography or sexually explicit content, and sexual content shared without the consent of the people in it or sent to people who did not ask for it.
Harassment — harassing, bullying, intimidating or stalking anyone.
Threats to life — threatening anyone's life or safety, or urging anyone to suicide or self-harm.
Hate — attacking people for their race, ethnicity, nationality, religion, gender, sexual orientation, disability or similar characteristics.
Violence — promoting or glorifying violence or terrorism, or showing graphic violence without context.
Impersonation — impersonating a person or organisation, or misleading people about who is speaking, including passing an assistant's words off as a person's.
Spam — spam, unsolicited bulk messages or outreach people did not agree to, and scams.
Private information — revealing other people's private information, such as their address, phone number or documents, or recording or transcribing people where the law requires their consent and they have not given it.
Malware and phishing — malware, phishing, or trying to break, overload or get around Spiral's security, its limits or the sandbox of interactive posts.
Nor may you try to make an assistant produce any of the above.
Everything in Spiral, interactive posts and the feed included, must be suitable for people aged 16 and over. Graphic content that breaks no rule — war news, injuries, medical images — is allowed in the feed; the filter or a moderator may cover it with "Show" (below).
Chats and assistants
People's messages in chats are end-to-end encrypted, so we cannot read or screen them; for those we rely on your reports (section 7). What assistants write is made on our servers: their replies in chats, the drafts they put to you for approval and the interactive posts they write pass automatic moderation (by OpenAI) before they are posted or shown to you, and what it flags is held back. If the moderation service cannot answer, the assistant's output is held back too.
How the feed is moderated
The feed is moderated by people, with automatic checks that help them. No AI reads or judges posts.
Who decides. Our moderator decides about posts, comments and accounts in a moderation queue. Every decision that restricts something is a person's. Automatic checks can only hold content back until a person looks at it.
Checks before a post or comment appears:
A keyword filter looks for a short list of severe terms in Ukrainian, Russian and English — slurs, threats, sexual content involving minors, other people's documents and card numbers, offers of drugs and weapons, and scams. It matches whole words and letters made to look alike. A match holds the post or comment: it is visible only to you, marked "Under review", until a moderator approves or removes it. Some terms — for example some wartime insults — and sexual language mark the post or comment sensitive instead (below). Ordinary swearing is allowed. The filter never tells anyone which word matched.
A link check refuses links to sites on public phishing and malware lists (CERT Polska's warning list and Phishing.Database) or on our own list. Nothing is fetched from a link and no preview is made.
The same post text posted twice by you within an hour is refused.
Limits: one post every 10 seconds and up to 5 comments in 30 seconds, and the daily limits of your level (below).
Trust levels. Your level comes from how you joined.
An account invited personally by someone with the standard level has the standard level at once: up to 200 posts and 2,000 comments a day.
An account that joined through one of our campaign links, or through an invitation from a new account, or without an invitation, has the new-account level for its first 3 days: up to 30 posts and 200 comments a day.
Links are allowed at every level; each one goes through the link check above.
Our moderator's and Spiral's own accounts have no limits.
A moderator can lower an account's level. That is a restriction: you get a notice saying why (below). Days are counted in UTC.
Names and bios. Your display name, profile photo and bio (if you set one) are shown with your posts. A name that matches our filter of severe terms is shown to others as "Spiral member"; a bio that matches is not shown.
What you see, and in which order. The feed shows every visible post, newest first. There is no ranking and no personalisation, with one exception: Spiral's own account writes its posts in several languages, and you see each in your app's language (English when it has none in yours). Above the newest posts there may be one pinned post from Spiral's own account, labelled "Official". Not shown: posts you hid, posts of people you block or who block you, and posts of accounts that are deactivated, locked or deleted.
Reports. Any account that is not suspended can report a post, a comment or a profile and name the rule it breaks (section 7).
A report for Child safety, Threats to life or Private information hides the post or comment from everyone at once, until a moderator decides, when it comes from an account whose reports count (below). Any such report alerts the moderator at once.
Otherwise, a post or comment is hidden once three different people with at least the standard level report it. Reports from new accounts, and from someone three of whose reports were found unfounded in the last 30 days, do not count toward the three.
A report still open after 24 hours hides the post or comment until a moderator decides, so a report never waits unanswered.
A report about a profile never hides anything by itself. A moderator decides about the person. These holds are automatic. The notices say so.
Decisions. A moderator can approve a held post or comment, remove it, mark it sensitive, or restore something removed by mistake. A moderator can suspend an account from posting and commenting for a period (reading and likes go on; a suspended account cannot report in the app and can write to Spiral Support instead, section 7), hold the posts it has at that time until a moderator approves them, or lower its level. A removed post stays visible to its author, marked "Removed by a moderator", until they delete it.
Notices. Every decision about your post, comment or account reaches you as a notice in the app (Feed → the bell → From Spiral) and as a notification: what was done and until when, what led to it (a report, several reports, the filter, a moderator, or a report open 24 hours), whether automatic means were used, the rule, the moderator's comment if there is one, and how to disagree. When the filter holds or covers your post or comment, it says so on the post and in a notice. A moderator checks everything the filter held or covered, usually within 24 hours. A notice never says who reported you. If you report something, you get a receipt with the time by which a moderator will decide, and the decision.
Sensitive posts. Graphic content that breaks no rule can be covered with "Show". Only the filter or a moderator marks a post or comment sensitive; you cannot. A sensitive post can be shared into a chat only after "Show", and the shared message starts with a warning. The filter also covers sexual language this way, at once; sexually explicit content still breaks the rules, and the moderator removes it when they check it.
Blocking works both ways in the feed. When you block someone, neither of you sees the other's posts and comments, and neither can like, comment on or share the other's posts. The block also applies to them everywhere in Spiral.
Your own posts. You can delete your posts and the comments under them, and turn comments off on a post.
If you disagree with a decision, write to Spiral Support at support@spiral.md with the notice. A person, not an automatic check, reviews it again, and we tell you the result. You can also go to court (section 13).
Authorities. When we suspect a crime that threatens someone's life or safety, we tell the police. We report child sexual abuse material to the competent authorities: Ukraine's Cyber Police and, where we can, the US National Center for Missing & Exploited Children. We keep the evidence for a year. We act on removal orders from public authorities.
Switching the feed off. We may make the feed read-only (nobody can post or comment) or switch it off for a time, for example while nobody can moderate it. Your chats are not affected.
If you break these rules we may remove the content, switch off your assistant, suspend posting in the feed, suspend or delete your account and stop you registering again — without notice when the case is serious. We report content to the authorities where the law requires it.
7. Reporting and blocking
In chats you can report a message (long-press it → Report content) or a room (the room's details → Report room), and block a person (their profile → Block user); a blocked person's messages and invitations no longer reach you.
In the feed you can report a post (the post's "…" → Report), a comment (its Report) or a profile (tap the author's name → "…" → Report), and block a person (tap the author's name → "…" → Block). You pick the rule and can add a note for the moderator; the author never sees the note or who reported. Up to 10 reports a day.
You can also write to Spiral Support, support@spiral.md — also if you have no account. Please send the link to the post (https://spiral.md/feed/…) or where the content is, why you think it breaks the law or these terms, your name and email address (not needed for child sexual abuse material), and that you believe the report is accurate and complete.
Reports reach us when you send them, and we act on them within 24 hours — in the feed a report still open then hides the content — and a person reviews each one. We remove content and accounts that break these rules; in the feed a moderator decides (section 6).
In an end-to-end encrypted chat we cannot read the message you report: we see who sent it, in which room and when, and the reason you give. We may ask you for a screenshot. Feed posts and comments are not encrypted: we see what you report.
8. Your content
You keep every right to what you write, say and upload.
You allow us to store, transmit and process your content only as far as needed to run Spiral for you and the people you talk to — including sending it to OpenAI or Google when you have allowed it (section 5).
In the feed you also allow us, free of charge and worldwide, to show your posts and comments, with your name, profile photo and bio, to every Spiral account in the order section 6 describes, and to keep the copies section 6 and the Privacy Policy say the moderation record needs. This permission ends when you delete the post or comment or your account, except for those records.
When you delete a post, its comments and likes are deleted with it. When a post is shared into a chat, the shared message is a copy sent by the person who shared it: it stays in that chat after the post or your account is deleted. A feed post can be shared only into Spiral chats, never into a chat bridged to another network.
We do not use your content to train AI models and do not sell it, and our AI providers do not use what we send them to train theirs (Privacy Policy, section 4).
You are responsible for what you send, including what you ask your assistant to write or publish, and for having the right to share it.
9. Open-source software
The Spiral iOS app is based on Element X (AGPL-3.0); Spiral's own code in it is licensed under the AGPL-3.0 or later. The app also contains other open-source components, listed in Settings → About → Open source licenses. Their licences give you rights these terms do not limit. The complete source code of each version is published at github.com/spiral-group/ios (Settings → About → Source code), and the Spiral authors also allow Spiral's own code to be distributed through app stores (the NOTICE file in the source says how). To request the source code, write to Spiral Support, support@spiral.md.
10. Ending your account
You can delete your account at any time in the app: Settings → Delete account. Deletion is permanent; the Privacy Policy says what is erased.
We may suspend or end your access if you break these terms, if the law requires it, or if we stop offering Spiral; where we can, we tell you first.
11. Availability and changes to Spiral
Spiral is free of charge and has no paid features. It is under active development: features may change, be limited or be removed. We try to keep Spiral running but do not promise that it is always available or free of errors.
12. Liability
Spiral is provided as it is. We take reasonable care to keep it working and secure, but as far as the law allows we make no promise that it will always be available, free of errors or fit for a particular purpose.
As far as the law allows, we are not liable for loss we could not reasonably foresee, for what other people or their assistants write, send or do, for decisions you make on the strength of what an AI wrote, or for interruptions caused by events beyond our reasonable control.
Nothing in these terms limits or excludes liability that the law does not allow to be limited or excluded — for example for death or personal injury caused by negligence, for fraud, or for damage caused on purpose or by gross negligence.
If you use Spiral as a consumer, the mandatory consumer protections of the country where you live still apply to you, and nothing in these terms takes them away.
13. Governing law and disputes
These terms are governed by the law of Ukraine, and the courts of Ukraine decide disputes about them. If you are a consumer living outside Ukraine, this choice does not take away the protection that the mandatory rules of your country's law give you, and you may also bring a claim before the courts where you live when that law allows it. Before going to court, please write to Spiral Support: most problems can be solved that way.
14. Changes to these terms
We may update these terms; the date at the top shows the version in force, and a new account accepts the version in force when it is created. If a change matters to you, we tell you before it applies — on this page and, where we can, in the app. If you do not accept the new version, you can stop using Spiral and delete your account (section 10).
Spiral Support is our single point of contact for people who use Spiral, and for the authorities of the EU Member States, the European Commission and the European Board for Digital Services (Digital Services Act, Articles 11 and 12), including removal orders under Regulation (EU) 2021/784. Write in English, Ukrainian or Russian.