Privacy Policy · Terms & Conditions · Delete account
Privacy Policy
This Privacy Policy explains what personal data the Argument mobile app and
API (“Argument”, the “Service”) collect,
why, who we share it with, and the rights you have. Argument is operated by
S O R S General Trading LLC, a company registered in the United Arab Emirates
(783315), Ras Al Khor Industrial Area 1, 10C street , Warehouse 4 (“we”, “us”). We are
the data controller for the personal data described here.
Effective date: 6 November 2026.
1. Who this applies to & minimum age
The Service is available worldwide. You must be at least 13 years
old to use Argument. The Service is not directed to children under 13
and we do not knowingly collect their data; if we learn we have, we delete it.
Where you are in the EU/EEA or UK and under the age of digital consent in your
country (up to 16), a parent or guardian must consent on your behalf.
Chats with other people are available only from the age of
16. The app asks you to confirm your age together with
the chat rules before your first chat, and we record when you confirmed
it.
2. Data we collect
Account & identity
- Email address and display name.
- A securely hashed password (bcrypt) — we never store your password
in plain text.
- If you use Google or Apple Sign-In, the identifier returned by that
provider (and, for Apple, the relay or real email you choose to share).
- An optional username you choose so people can find you
in Chats. Your username and display name are visible to other users who look
you up or chat with you.
- An optional profile photo. Our server re-saves it as a
small square picture (512×512 pixels) without its metadata (such as
location or camera details) and screens it automatically, like photos in
chats, before anyone can see it; a photo the screening flags is not kept.
By default only the people you chat with see it — people you wrote
to or whose message request you accepted. In Settings you can choose to
show it to everyone who looks you up instead; switching back makes the
links other people already received stop working. It is never shown
between you and someone either of you has blocked. Replacing or removing
it deletes the previous photo.
- Which version of our Terms and this Privacy Policy you accepted, and
when.
Verification & account security
- One-time email verification codes (6 digits), automatically deleted
shortly after they expire (within roughly one hour). A record that a code
was sent to your address (without the code) is kept for a day, to limit
how many can be requested.
- Session tokens (signed JWTs) issued when you sign in.
- Anti-fraud signals to prevent trial/subscription abuse: a one-way
SHA-256 device fingerprint, optional device model/OS/screen metadata, and a
normalized form of your email address used only to detect duplicate
accounts.
Argument content
- The text you submit during an argument, transcripts of any voice input,
the language, and the AI-generated titles, analysis and follow-up answers.
- Voice input is ephemeral. When you speak your side of
an argument, the audio is sent for transcription and then
permanently deleted from our servers and from the AI
provider’s file storage immediately afterwards. We may keep a derived
emotion label (calm / neutral / upset) inferred from tone — not the
audio itself.
- Live recordings stay on your phone. When you record a
conversation with Live, the recording is saved only in the app’s
private storage on your device (excluded from device backups) and is
deleted together with its result or when you delete it. While you record,
the app sends short parts of the audio (about 20–40 seconds each),
together with up to four short samples of the voices already heard (so each
speaker keeps the same label), to our servers, which pass them straight on
to OpenAI for transcription with speaker separation. We do not store
this audio. We keep the resulting transcript (text, timing and
“Voice 1, Voice 2…” labels), the names you give
the voices and the corrections you make. Speaker separation only tells the
voices in one recording apart; it is not used to identify anyone —
except you, if you turn on “My voice” (below).
- “My voice” (optional, off by default): if
you turn it on in Profile, the app records a short sample of your own voice
(about 10 seconds of you reading a sentence) and keeps it only on your
phone, in the app’s private storage (excluded from device backups).
With every part of your Live recordings the app sends this sample along
with the audio, through our servers, to OpenAI, so the speaker separation
marks your lines as yours. We do not store the sample;
OpenAI processes it like the recording and does not use it to train its
models. We keep only the fact that you turned it on, when, and which
version of the consent text you agreed to, and a recording’s consent
record notes that your sample was used. It is used only to recognise you,
never anyone else. Turn it off in Profile → My voice at any time: the
sample is deleted from your phone and the record from our servers; it is
also deleted when you sign out or delete your account.
- Proof of consent for Live: Argument does not notify
the other people present — telling them is up to you. Before each
recording you confirm that you have told everyone present and that they all
agree to be recorded. We keep that confirmation with the recording’s
result — when you confirmed it, which version of the consent text you
confirmed, and your app version and platform (for recordings made with
earlier versions of the app, also whether its spoken announcement played)
— and it is deleted together with the recording.
- Post and fact checks: the link or text you check, your
objection, and the result with its sources. A screenshot you check is sent
to the AI for that check only and is never stored.
Chats
- Messages you send and receive, the photos, PDF files and voice
messages you attach, when they were sent and read, who you chat with,
message requests, and your mute and block settings.
- Voice messages: the audio is stored on our servers so
it can be delivered and played. Before it is delivered, it is sent to
Google’s Gemini API to be transcribed; the transcript is screened like
a text message, is stored with the message, is shown to the people in the
chat when they tap “Show text”, and is what Argument reads when
it replies to a voice message or when a dispute in the chat is settled.
Voice messages can only be sent in chats that both people have
accepted.
- Photos are re-saved on your phone before they are sent, which removes
location and camera details (EXIF).
- Messages and files are stored on our servers so they can be delivered
and shown on your devices. They are encrypted in transit and at rest, but
chats are not end-to-end encrypted.
- Automated safety screening: before a message or photo
is delivered, it is checked automatically for illegal or abusive content
(see Section 4). Content that breaks our rules is not delivered.
- Reports and automated decisions: when someone reports a
message or a person, we keep a copy of what was reported (the message, or
that person’s recent messages to the reporter, with their photos).
The report is reviewed automatically by our AI provider against our chat
rules (see Section 4). When the breach is clear, the message is removed
for everyone and the sender’s chats are closed for 7 days, or
permanently after a repeated breach; unclear cases, and reports from
someone the reported person had blocked, go to a human moderator. If three
different people report someone within 7 days, that person’s chats are
closed until a moderator reviews the reports. We keep a log of each
decision (what was decided, when and why — without the messages
themselves). You can ask a person to review any such decision (see
Section 3).
- Suspected child sexual abuse material: if our
automated screening flags a message or photo as sexual content involving
minors, it is not delivered, the photo is deleted, the sender’s chats
are closed until a moderator reviews it, and we keep an incident record:
the account, the time, the categories and a cryptographic fingerprint
(SHA-256) of the content — never the content itself. Where the law
requires, we report such incidents to the competent authorities.
- Our moderators are told by email about cases that need a person,
appeals and urgent incidents; these emails never contain messages or
photos.
- Messages and photos you send in the chat with Argument are processed by
our AI providers to write its replies. When you ask it to check a fact, or
a reply needs a fact checked, the claim is checked against web sources like
any other fact check and the check is kept in your History.
- Settling a dispute in a chat with a person: nothing in
a chat between people is analysed unless one of you asks Argument to
settle a dispute. Then up to the last 100 messages of that chat that you
both can see (voice messages as their transcripts, photos only as a mark
that a photo was sent) are sent to our AI provider to find where the
current dispute starts and ends; nothing is saved from that search. You see
the stretch it found and can change where it starts; the other person sees
the same stretch and agrees or declines. The verdict is given only on the
stretch of messages you both agreed to, and it is shown to both of
you.
Calls with Argument
- In the chat with Argument you can call Argument and talk out loud.
During a call your phone sends your voice in real time directly to
OpenAI, whose GPT-Live model hears it and answers in
Argument’s voice. The audio of a call does not go through
our servers and we do not record or store it.
- To start a call, our servers give OpenAI Argument’s instructions,
what it remembers about you (if memory is on) and the latest messages of
your chat with Argument. When you ask it to check a fact or look something
up during the call, OpenAI may pass your question to another of its models,
which can search the web.
- After the call your phone sends us the text of what you and Argument
said. We keep it with the call so you can read it again; the chat shows
only a card with the call’s length. You can delete the text of any
call (“Delete the text”); the card stays. We also keep when each
call took place, how long it lasted and how it ended, to count the call
minutes of your plan.
- If memory is on, Argument remembers from a call what it would from the
chat (see below).
Argument’s memory
- In your chat with Argument, Argument keeps a short summary of earlier
topics and a limited number of facts you have told it about yourself (for
example your name, how you like to be addressed or what you are working
on), so that its replies can take them into account.
- It never keeps facts about your health, sexual orientation, religion,
political views, precise location, contact details or finances, and about
other people at most a first name and how they relate to you. Email
addresses and phone or card-like numbers are removed before anything is
saved.
- You can see everything it remembers, delete any fact or all of them,
and turn memory off in Profile → Memory or from the chat menu. With
memory off, Argument neither saves new facts nor uses the saved ones.
Memory is never used when Argument judges a dispute between people.
Your portrait
- After 3 finished disputes, Argument writes a short portrait of how you
argue: a type, four traits, a strength, what to work on and a few tips.
It is available on every plan and shown only to you.
- For this, we keep a short note about each dispute you finish
(including Live recordings and disputes settled in chats), about
your own side only: the share of the verdict you received,
whether your side held up, the scores the verdict gave you where it
measured them, the judge’s remarks about your side and your strongest
points — never the other people’s words. These notes are kept
even on plans that do not keep a History.
- The portrait is written by our AI provider from these notes, in your
app’s language. It is an automated observation for your own
reflection, not a diagnosis or a judgement of you as a person, and it is
never used to make decisions about you or your account.
Connected AI assistants
- If you connect an AI assistant such as ChatGPT or Claude (Pro plan), we
keep a record of the connection (the assistant’s name, when it was
connected and last used) and access tokens stored only in hashed form.
- When you ask it to, the assistant can read your arguments (your own
statements, the analyses and verdicts, and transcripts of Live recordings you
made), your statistics and your saved checks. It cannot read your chats or
other participants’ written statements, and it cannot change
anything.
- What the assistant reads is then handled by its provider (for example
OpenAI or Anthropic) under that provider’s own terms and privacy
policy. You can disconnect an assistant at any time in Profile → AI
assistants; its access stops immediately.
Social & safety
- Participants and invitations for arguments you join or create, and your
in-app contacts (people you have argued with).
- Reports you file and feedback you give on AI results. When you report
a message or a person, we keep a copy of the reported message (or of that
person’s recent messages to you) so our moderators can review it, even
if it is deleted later.
Purchases
- Subscription and one-time-pass records: the store (Apple App Store or
Google Play), product ID, transaction identifiers, purchase token,
subscription status and expiry. We do not receive or store your
card or bank details — Apple and Google process all
payments.
Notifications
- If push notifications are enabled, a device push token, the platform
and the app’s language so we can notify you about invites, new
messages and completed analyses. Chat notifications only say “New
message”; they never include the message text.
Analytics & diagnostics
- Crash & error diagnostics (via Sentry): when the app
or our servers encounter an error, we collect a crash/error report —
device model, OS version, app version, a stack trace, and a pseudonymous
account identifier. Email addresses and authentication tokens are scrubbed
before the report is sent; we do not collect your argument content here.
- Product analytics (via PostHog): app-interaction events
(e.g. screen views and actions such as “argument created”), an
approximate region derived from your device language setting (not precise
location), and a pseudonymous account identifier. We do not
record session replays, and event properties never include your argument
text, transcripts, or email.
- Ad conversion measurement (via Google Analytics for
Firebase): with the same consent as product analytics, we record standard
app events (such as first app open, sign-up, and purchase) so Google can
tell us whether our own advertising campaigns lead to installs and
purchases. On Android this may use your device’s advertising ID; on
iOS we do not use the advertising identifier. This data is never used to
show you personalized ads and never includes your argument text,
transcripts, or email.
- Your choice: product analytics and ad-conversion
measurement are opt-in where required (EEA/UK) and can be turned off at any
time under Profile → Analytics. Crash/error diagnostics run to keep
the Service stable and secure.
3. How we use your data and our legal bases
- To provide the Service — host arguments, generate
analysis, manage your account and subscription (legal basis: performance of
our contract with you).
- Security & abuse prevention — authentication,
anti-fraud, handling reports and enforcing our Terms (legal basis: our
legitimate interests in keeping the Service safe).
- Communications — email verification, password
reset and essential service messages (legal basis: contract).
- Keeping chats safe — automated screening of
messages, photos and profile photos, automated and human review of reports, closing the
chats of accounts that break our rules and keeping incident records (legal
basis: our contract with you, our legitimate interests in a safe Service,
and our legal obligations).
- Argument’s memory and your portrait —
making Argument’s replies and your portrait personal to you (legal
basis: performance of our contract with you; you can turn memory off,
delete what it keeps or reset the portrait at any time).
- Records of consent and acceptance — keeping proof
that the people in a Live recording agreed to it, which version of our
Terms and this Policy you accepted, and your confirmation of the age for
Chats (legal basis: our legitimate interests in being able to show that
the Service is used lawfully, and our legal obligations).
- Stability & security diagnostics — crash and
error reporting to detect and fix faults (legal basis: our legitimate
interests in a stable, secure Service).
- Product analytics & ad measurement —
understanding which features are used and whether our advertising brings
new users, so we can improve and grow the Service (legal basis: your
consent where required, e.g. EEA/UK; otherwise our legitimate interests,
with an opt-out).
- Legal compliance — meeting obligations under
applicable law (legal basis: legal obligation).
Automated decisions (GDPR Article 22). Removing a
reported message and closing someone’s chats can be decided
automatically, because reports must be handled quickly to keep Chats safe;
this is necessary for performing our contract with you. You always have the
right to have a person review such a decision, to express your point of view
and to contest it: tap “Ask a person to review” in the app, or
email support@argumentapp.ai. A moderator
then reviews the decision and can reverse it, which also restores removed
messages.
For users in the UAE, processing is carried out in accordance with Federal
Decree-Law No. 45 of 2021 on the Protection of Personal Data. Where we rely on
consent (e.g. optional notifications) you may withdraw it at any time.
4. AI processing
To produce your result, the text you submit and any transcript are
processed by Google’s Gemini models: the reasoning level you choose
(Standard, Advanced or Maximum, depending on your plan) decides which Gemini
model writes a verdict or Argument’s reply. Analyses, Live hints, post and fact
checks (including those in the chat with Argument), Argument’s replies
in Chats and its memory, finding the dispute to settle in a chat, your
portrait and translations are routed primarily through
OpenRouter (OpenRouter, Inc.), an AI request router that
forwards them to the same Gemini models, with Google’s Gemini API used
directly as a fallback. Voice input for arguments is sent directly to
Google’s Gemini API for transcription and deleted immediately
afterwards; voice messages in chats are sent directly to Google’s
Gemini API for transcription (see Section 2). For post and fact checks
the AI model searches the web for sources. Copies of reported chat content
are processed by the same Gemini model, through the same routes, to review
reports against our chat rules.
OpenAI (OpenAI, L.L.C.) runs calls with Argument: its
GPT-Live model hears your voice and answers out loud in real time, and
another of its models may search the web for a question you ask during the
call (see Section 2). OpenAI also processes the audio of Live
recordings (with your own voice sample if you turned on “My
voice”), for transcription with speaker separation; chat messages,
photos and profile photos, for automated safety screening before anyone sees
them; and, only if the
other routes fail, translations of results. OpenAI does not use this data to
train its models; under its API terms it may keep it for up to 30 days to
detect abuse, after which it is deleted.
Your content is processed solely to provide the Service and is
not used by us to train AI models, and not sold. Use of these
providers is subject to their own terms and privacy policies.
5. Sharing & sub-processors
We do not sell your personal data and we do not show third-party ads in
the app. With your consent, we use Google Analytics for Firebase solely to
measure whether our own advertising campaigns lead to installs and purchases
— never to build advertising profiles or personalize ads. We share data
only with service providers acting on our instructions:
- Google — Gemini AI processing (including web
search for checks and the transcription of voice messages), Google Play
billing verification, and (if enabled) push notification delivery.
- OpenRouter, Inc. — primary routing of text AI
requests to Google’s Gemini models.
- OpenAI — calls with Argument (your voice in real
time, and web search for questions asked during a call), transcription of
Live recordings (and your “My voice” sample, if on),
automated safety screening of chat messages, photos and profile photos,
and a last-resort fallback for translating results.
- Cloudflare (R2 storage) — storage of the
photos, files and voice messages sent in chats, and of profile
photos.
- Apple — Sign in with Apple and App Store billing
verification.
- Resend — delivery of verification and
password-reset emails.
- Render — cloud hosting of the Service.
- Sentry (operated by Functional Software, Inc.) —
crash and error diagnostics, hosted in the EU region.
- PostHog — product analytics, hosted on
PostHog EU Cloud.
- Google Analytics for Firebase — consent-gated
app analytics used solely to measure our own advertising campaigns.
The people you chat with receive the messages and files you send them.
An AI assistant you connect receives what it reads on your behalf; it acts
under its provider’s own terms, not on our instructions. We may also
disclose data where required by law, to protect our rights, or in connection
with a corporate transaction.
6. International transfers
We operate globally and our providers (AI, email, hosting) may process data
in countries outside your own, including outside the UAE and EEA. Crash
diagnostics and PostHog product analytics are hosted in the European
Union; Google Analytics for Firebase data may be processed on Google
servers worldwide, including in the United States. Where we transfer personal
data internationally we rely on appropriate safeguards such as standard
contractual clauses or equivalent mechanisms.
7. Retention
- Email verification codes: deleted shortly after expiry (~1 hour); the
record that one was sent, within a day.
- Voice input audio: never stored (deleted immediately after
transcription).
- Live audio: kept only on your phone and deleted with its result or when
you delete it; our servers never store it. Transcripts are kept with the
result.
- Argument content, Live transcripts and saved checks: kept until you
delete the relevant history item or your account.
- Chat messages, files and voice messages (with their transcripts): kept
until you delete them for everyone or delete your account. Deleting a conversation removes it from your list only;
the other person keeps their copy. Uploads that were never sent are deleted
within a day.
- A message you delete for everyone in a chat with another person
disappears from the chat at once. Its text and files are held, shown to
nobody, for up to 7 days so that a report about it can still be reviewed
(longer only while such a report is open), then deleted.
- Profile photo: kept until you replace or remove it or delete your
account; the previous photo is deleted at once.
- Reports and the copies of messages they keep: until the report is
decided and then for 12 months, after which the copies are deleted. The log
of moderation decisions is kept for 12 months.
- Photos of messages removed by moderation, or deleted by their sender
while a report about them is open: held, never shown, for up to 30 days
(longer while the report is still open) so a moderator can review them
and an appeal can restore them; then deleted.
- Records of chats closed by moderation: kept while your account
exists, so that a repeated breach can be recognised.
- Incident records of suspected child sexual abuse material (never the
content): kept as long as necessary to meet our legal obligations.
- Proof of consent for a Live recording: kept with the recording’s
result and deleted with it.
- Calls with Argument: we never store their audio. The text of a call
is kept until you delete it or your account; when each call took place,
how long it lasted and its card in the chat are kept until you delete your
account.
- Argument’s memory (facts and topic summaries): kept until you
delete them or your account.
- Portrait notes and the portrait: kept until you reset the portrait or
delete your account; hiding a dispute from your History deletes its note
and the portrait is written again without it.
- Your acceptance of our Terms and this Policy (version and time) and
your confirmation of the age for Chats: kept while your account
exists.
- Connected assistants: access tokens expire after an hour and renewal
tokens after 60 days without use; all are deleted when you disconnect the
assistant or delete your account.
- Subscription records: kept while your entitlement is active and as needed
for billing/fraud and legal reasons.
- Anti-fraud identifiers (one-way hashed device identifier, normalized
email): retained after account deletion, solely to prevent trial and
sign-up abuse, including signing up again to get around chats closed by
moderation.
- Backups containing prior data are rotated out within
30 days.
8. Your rights
Depending on where you live, you may have the right to access, correct,
delete, export, restrict or object to processing of your personal data, and to
withdraw consent.
- EU/EEA & UK (GDPR / UK GDPR): all of the above, plus
the right to lodge a complaint with your data protection authority (in the UK,
the ICO), and the right not to be subject to a decision based solely on
automated processing without human review (see Section 3).
- California (CCPA/CPRA): the right to know, delete and
correct, and to opt out of “sale” or “sharing” —
we do not sell or share personal data as defined by that law — with no
discrimination for exercising your rights.
- UAE (PDPL): equivalent access, correction, deletion and
objection rights.
You can exercise most rights directly in the app, or contact us at
support@argumentapp.ai.
9. Deleting your account
You can delete your account at any time: open the app, tap the profile
icon, then Delete account. When you do, your email, display
name, password and any Google/Apple links are immediately anonymized and the
account is deactivated. To keep the record intact for the people you argued
with, your participation in past arguments is retained in anonymized form
(you appear as “Deleted user”). The text of messages you sent in
chats is erased (people you wrote to see “Message deleted”), the
photos, files and voice messages you sent are deleted, your username is
released, your profile photo is deleted, your chat with Argument, your calls with it (with their texts) and
everything it remembers are deleted,
your portrait and the notes it was written from are deleted, connected AI
assistants are disconnected and your push tokens are removed. To prevent abuse (such as
repeated free trials or duplicate accounts) we also retain a one-way hashed
device identifier and a normalized form of your email address; these
anti-fraud records are used only to detect duplicate or abusive sign-ups,
are never used to contact you, and serve no other purpose. Records kept for
safety and legal reasons (reports and moderation decisions, records of
closed chats, incident records) are kept for the periods in
Section 7. Backups are rotated out within 30 days. See the
account deletion page for full details,
including what to do if you have lost access.
10. Security
We protect data with measures including bcrypt password hashing, signed
session tokens with a limited lifetime, encryption in transit (HTTPS/TLS),
encrypted storage, and ephemeral handling of voice audio. Chats are not
end-to-end encrypted. No system is perfectly secure, but we work
to protect your data and to notify you and regulators of incidents where
legally required.
11. Changes
We may update this policy. Material changes are reflected by updating the
“Last updated” date, and the app asks you to accept the new
version before you continue; we record which version you accepted and
when.
12. Contact
S O R S General Trading LLC, Ras Al Khor Industrial Area 1, 10C street , Warehouse 4, United Arab Emirates.
Privacy & data requests:
support@argumentapp.ai.
Argument · operated by S O R S General Trading LLC · argumentapp.ai · Last updated: 6 November 2026