Skip to content
VANTYX

VANTYX privacy policy

DRAFT. NOT LEGAL ADVICE. NOT REVIEWED BY A LAWYER.

This was written by the people who built VANTYX, not by a solicitor. Neither the author nor the owner of VANTYX is a lawyer, and nothing here is legal advice. It is published as a draft so that what the software actually does is on the record in plain language while the real document is being prepared. It must be reviewed by a qualified solicitor before it is relied on by anyone. Nothing here has been reviewed. Where this draft and the software disagree, the software is what is true and this page is what is wrong.

Who this is for

You run a FiveM server. VANTYX is an anti-cheat that runs on it. This page is written for you, in the same plain language the product uses, and it is checkable — every claim below corresponds to code in the VANTYX repository, and the technical version of this document names the file.

There are two kinds of data here and they are worth keeping apart. There is your data: your email address, your password, your licence, your servers. And there is your players' data: who connected, what VANTYX observed, what it flagged. You decide what happens to the second kind. VANTYX handles it for you and on your instructions.

The short version

VANTYX does not fingerprint hardware. It does not capture screens. It does not scan memory or processes. It does not read files on a player's computer. It does not share bans between customers. And it never bans anyone by itself.

Those are unusual claims in this market, and they are the reason this page is worth reading rather than skipping. Each of them is a design constraint with code behind it, not a promise, and each of them costs us something — which is stated further down, because a refusal with no cost attached is marketing.

There is nothing on your players' computers

VANTYX is a server-side FiveM resource. It installs on your machine and runs there. Your players download nothing, install nothing and run nothing from us. There is no VANTYX process, service, driver or agent on any player's computer, and there is no code path that could put one there.

This is why the list of things VANTYX cannot see is long. Everything it knows, it knows because your own server already knew it.

What leaves your machine, and what does not

The most important sentence on this page: the parts of VANTYX that see the most see it on your server and never send it anywhere.

Your players' positions, health, speed, weapons and actions are sampled continuously on your own machine, checked against the rules there, and thrown away. What travels to VANTYX is a finding — "implied speed was 412 metres per second where the ceiling for this vehicle class is 90" — or a pre-aggregated five-minute counter. It is never a movement track, never a recording, never a log of everything that happened.

The uplink is deliberately small: roughly two kilobytes per second plus twelve bytes per player slot. That is a budget your server enforces on itself. It is not enough to stream raw behaviour even if somebody wanted to, which is the point.

Your players' IP addresses are cut down before they leave your server. VANTYX masks an IPv4 address to its /24 network and an IPv6 address to its /48 — so 203.0.113.47 becomes 203.0.113.0/24 — inside the resource running on your machine, before the value is put into anything that gets sent. VANTYX never receives a full player IP address. If VANTYX were breached tomorrow, there would be no full player IP address in it to take, because one was never sent.

What VANTYX holds about your players

Account identifiers. The Rockstar licence, Steam, Discord, Cfx.re, Xbox Live and Microsoft account identifiers that FiveM already reports to your server for every player. These are stored as one-way hashes, computed with a key that is unique to your organisation and that never leaves our key-management service. The practical effect is that the same player produces different values in your data and in another customer's, so the two cannot be joined — and a stolen copy of the database yields no identifiers at all, because the key is not in it.

Alongside each we keep the last six characters of the original so your staff can recognise a player they are looking at. For IP addresses and Cfx tokens even that is withheld, and the database refuses to store it.

One honest correction about the current build. The hashing above is written, tested and is the design — and it is not yet on a live code path. Today those identifiers travel from your server to us, sit briefly in memory and in a processing queue, and are discarded rather than being written to the database at all. That is less data than this page describes, and it is also less tidy than it describes, so it is being fixed rather than reworded. Nothing else in this section is affected.

A Cfx entitlement token, and what it is not. FiveM exposes a token tied to a Cfx.re account. We store up to four of them, hashed. It is not a hardware identifier, it rotates, and it collides between people who share a machine or use cloud gaming — so VANTYX treats it as weak evidence and the database physically refuses to let it be the only basis for a ban.

A network prefix, as described above. Never a full address. It is used to notice when several players share one connection, so that a household, a student hall or a phone hotspot is not banned as though it were one cheater.

The in-game name your player chose, so your staff know who they are looking at. It decides nothing.

Connection quality — ping, jitter, packet loss, country. This is mostly used against detections rather than for them: a bad connection can imitate several kinds of cheating, and VANTYX needs to know when a measurement cannot be trusted.

What VANTYX found. Which check fired, the numbers behind it, the plain-English explanation, the legitimate causes VANTYX considered, and the running risk score. This is the product.

Enforcement records. Bans your staff issued, the evidence bundle behind each one, appeals, and your staff's own notes.

What VANTYX will never collect

Not "does not currently". These are refusals built into the system, and the software rejects the data rather than choosing not to ask for it.

No hardware fingerprinting, no HWID. There is no identifier type for one. The list of identifier types VANTYX understands is fixed and closed, and none of them describes a device.

No screen capture. No screenshots, ever, by any means.

No memory or process scanning. VANTYX never reads a player's memory or lists their running programs. It cannot: there is nothing of ours on their machine to do it with.

No file listings from a player's computer. No paths, no directory scans, no installed-software inventory.

No window titles, no clipboard, no browser data.

No cross-customer ban sharing. There is no shared ban network and there is no way to build one by accident: because each customer's identifiers are hashed with a different key, two customers' records cannot be joined. If we ever build a threat-intelligence feature it will exchange counts and never rows, and it will be off unless you switch it on.

No automated bans. VANTYX's ceiling is proposing a temporary ban. A named human on your staff decides, and the database will not accept a ban row without an evidence bundle attached, or a permanent ban without a person's name on it. This is enforced in the database, not by policy, which means it cannot be turned off by a configuration change or a bug in application code.

What each refusal costs you

Stated because you would find out on your own server anyway.

Because there is no hardware identifier, ban evasion is cheap. Every identifier we can hold rotates or collides. A determined cheater can come back with a new account. VANTYX detects linked accounts as a signal for your staff, not as an automatic ban.

Because there is nothing on the player's machine, an external cheat that only reads memory is invisible. ESP and radar tools that never send a false value to your server cannot be seen by any server-side anti-cheat, including this one. VANTYX catches what a cheat does to your server, not what it displays on a player's monitor.

Because a backdoored resource on your server runs with the same privileges VANTYX does, VANTYX can flag it and show it as a risk, but it cannot contain it.

Artificial intelligence is off unless you turn it on

By default, VANTYX writes its own explanations. The plain-English account of why a check fired, what it means, and what might legitimately explain it is produced by VANTYX itself from its detection catalog and the recorded evidence. Nothing about your server or your players is sent to any AI provider. This is the default on every VANTYX deployment, and it is not a degraded mode — it answers every question the AI version answers.

A second honest correction: in this build you cannot switch it on. The machinery is complete — an owner-only permission, a disclosure to read and acknowledge, somewhere to store your own Anthropic key sealed so that a copy of our database is not a copy of your key, a spend cap and a stop button — and none of it is reachable, because the dashboard page that would show it has not been built. So the paragraph below describes what happens when you enable it, and today nobody can. We would rather say that than let a policy page imply a choice you do not yet have.

If you switch the AI Investigator on, high-risk cases additionally get a written analysis from Anthropic's Claude. Enabling it means giving us your own Anthropic API key, which we store encrypted and never show or send back to you — only a short fingerprint, so you can tell which key is loaded. When an analysis runs:

by us. The provider cannot reverse it, and the same player has a different pseudonym in another customer's data — and even on another one of your own servers.

included. Not filtered out — never assembled in the first place. The payload is built field by field from a fixed list, so a new piece of data added elsewhere in the system cannot leak into it.

resembling an address, an identifier or an email. It does not quietly remove it; it stops and falls back to the non-AI explanation.

The AI never decides anything. It writes; it cannot move a risk score, propose a ban or change an outcome.

What VANTYX reads from the other resources on your server

Only the manifest. For each resource on your server, VANTYX reads fxmanifest.lua — the small declaration file, not the code — hashes it, and records the resource name, whether it is running, and that hash. It does not read scripts, assets, models or configuration belonging to any other resource.

For weapon and vehicle values, VANTYX prefers what you or a resource declares through its integration API or uploads through the dashboard. In this version it parses no .meta file belonging to another resource at all; the base-game values it uses are compiled into VANTYX. If that ever changes, it will read numeric fields only, will store a hash and a version rather than the file, and this page will say so.

There is a separate document for asset vendors who want the detail.

Your own account data

For your account, VANTYX is the one making the decisions, so this part is about us.

We hold your email address, your display name, a hash of your password, and — if you use it — an encrypted second-factor secret. Passwords are stored with argon2id and are never recoverable by anyone, including us. We hold your organisation's name, billing email, plan and licence state.

We do not hold your card details. Payment is handled by Stripe and card data never passes through VANTYX; we keep only the reference numbers needed to know which subscription belongs to which account.

We keep an audit log of what your staff did in the product — who banned whom, who changed which setting, from which network. That log is append-only: it cannot be edited or deleted by us or by you. It exists to protect the people your staff make decisions about.

How long things are kept

These periods are the same for every customer on every plan. They are set in one place in the code and the cleanup jobs read them from there.

and a meaningless label is left in place

the day it was issued, because that is how long a dispute about it can still arrive

Evidence that supports a ban currently in force or an appeal currently open is not stripped while that is true. A player disputing a ban has to be able to see the case against them.

Deleting a player's data

Some of what VANTYX holds cannot be deleted, and we would rather explain why than be vague about it. The evidence records and the audit log are append-only by design: a record of a decision made about a person is worth nothing if it can be quietly rewritten afterwards.

So the system was built to avoid the conflict rather than to argue about it. Everything that identifies a human is kept separately from the evidence itself, in a table designed to be deleted from. The evidence record and its tamper-evident chain contain no personal data at all — this is checked on every single write, and a check that fires refuses the write. Erasing a person therefore means deleting the personal parts and leaving a record that no longer refers to anybody.

Be aware of the honest limit: what remains is pseudonymised, not anonymous. The history is severed from the identifiers, not from the fact that some player did those things on your server.

Erasing a person is now a single request rather than an operator running SQL by hand. The response is not a confirmation — it is an itemised account of what was destroyed and what remains, with the lawful basis for each thing that remains, because "we deleted everything" would not be true and a person asking is entitled to the accurate version.

What is destroyed: every identifier we held, deleted outright rather than hidden, so the account cannot be recognised again; the in-game name everywhere it was stored; the network the person connected from; and staff notes, redacted.

What remains: the flags and the evidence behind them, attached to a random internal number and to no person, because the records are chained and removing one would break the proof that the others have not been altered. The record that staff acted, which never contained a name. And if a ban was issued, the ban itself and the unreadable keyed values it is enforced on — otherwise asking to be erased would be a way of asking to be unbanned.

Two limits stated plainly. If a ban exists, those keyed values stay for as long as the ban record does, which for a permanent ban is up to seven years; the request is recorded as complete as to identity and not as complete overall, and that distinction is deliberate. And point-in-time backups are not rewritten — they are encrypted, never restored selectively, and expire on their own schedule, after which no copy remains anywhere.

There is still no button in the dashboard. The customer's staff make the request through the API, and if you have received a request from a player and are not sure how, contact us.

Where the data lives

One region: AWS in Frankfurt (eu-central-1). Backups are held there too.

Three companies can reach parts of it: Amazon Web Services, which hosts everything; Stripe, which handles subscriptions and sees only your billing details and never any player data; and Anthropic, which sees pseudonymous case data and only while you have the AI Investigator switched on.

Bans and appeals

Every ban VANTYX carries comes with an appeal code, shown to the player at the moment they are removed. With that code, and without any account, they can read the evidence behind their ban in plain language and write a reply. Your staff decide the appeal — VANTYX only carries it.

The player is not shown the exact measurements, the thresholds those measurements crossed, the internal explanation of how the check works, or anything about any other player. Publishing exact thresholds to a banned cheater is publishing the manual for getting away with it next time. The page says that plainly rather than presenting a partial record as a complete one.

There is a separate document describing the appeals process in full.

Things this draft does not yet answer

Left here deliberately, because a policy page that quietly omits its open questions is worse than one that lists them.

and VANTYX has no way of knowing a player's age.

entity that would receive the mail has not been decided.

Changes

This is a draft. It will be replaced by a version a solicitor has written. When that happens, this notice goes away and the document gets a date and a version.