Privacy
Plain language. This page describes everything the realm stores about you and how it is used. No legalese, no third-party data sale, no analytics or trackers of any kind. The realm is designed to run unattended — every rule below is enforced by the code, not by a human moderator on call.
What we store about you
About your IP address
About your email address
What an administrator will never ask you
Deleting your account
Security
Automatic chat moderation
The realm runs an automatic moderation system on the public chat channels (General Trade LookingForGroup GuildRecruitment World /say /yell). It never reads private channels — no whispers, no party chat, no guild chat, no custom player-made channels — those are off-limits by design.
The watcher is an AI tool running on the realm's own machine — every public-chat line stays on the realm, nothing is sent to any outside service. It sorts what it sees into four severity levels. Casual swearing, game slang and venting are not sanctioned — only repeated personal insults, hate speech and severe content are.
Severity scale
| Level | What it covers | Score cost |
|---|---|---|
| 0 — OK | Normal chat. Includes frustrated swearing, casual slang (FR "putain", "wesh"; EN "wtf", "lol", "noob"), game jargon, trade calls. | 0 pts |
| 1 — Mild | Light personal insult directed at another player. | −10 pts |
| 2 — Serious | Sexist, racist, homophobic or xenophobic content. Repeated harassment. Clear hate speech. | −20 pts |
| 3 — Severe | Death threats. Slurs explicitly targeting a person. Doxxing. Sexual content involving minors. Gold-seller advertising URL. | −30 pts + immediate 30 minutes mute |
Good-conduct score
Every account starts at 100 points. Each infraction subtracts the score cost shown above. The score regenerates at 10 points per day automatically — meaning a single bad moment costs you almost nothing in the long run as long as it stays isolated. With the realm's current values, an account would need roughly 10 consecutive level-1 infractions (or 5 level-2, or 4 level-3) to reach zero score and trigger the first mute.
When the score hits zero
An automatic chat mute of 1 hour applies right away. You can still log in and play; you just cannot post in public chat for that duration. Once the mute fires, the score is restored to half-maximum (100/2 = 50 points by default) so a chatty player coming out of a mute has enough margin to participate without immediately tripping another sanction on a single later infraction. The mute is the actual punishment; the partial restore is a fresh-start gesture, not a free pass.
Recidivism
After a mute fires, the account is flagged as "recently sanctioned" for 3 days. If the score reaches zero again while the flag is active, the next sanction is harder: a 1 day mute. The flag clears automatically after 3 days of clean behavior.
Permanent ban after repeated recidivism
The system counts repeat-offense sanctions across the account's entire lifetime (this counter is not cleared by the 3 days window above — that window only governs how severe the next sanction will be, not the counter itself). When this counter reaches 5, the next sanction is no longer a temporary mute but a permanent account ban. This is the final step, reached only by accounts that keep going back to the same behavior despite warnings and mutes. The counter is visible to the operator from the admin console.
What we keep and for how long
Chat lines from the watched channels sit in a short-lived holding area while they wait to be checked. Once a line is checked it is deleted; lines still unchecked after 6 hours are deleted unchecked. No public-channel message is kept word-for-word past that window. What we do keep, per account: the current score, the recidivism flag, the date of the last event, and a short one-sentence verdict for the most recent infraction (so you can see why you were warned). Sanctions are kept in a history for the operator; older entries can be cleared out.
Margin for error
This realm is designed to run without live moderation — there is no GM ticket queue, no Discord, no support email. False positives still happen (especially on casual French banter early on), but the sanction sizes are calibrated for that: even the longest first-time mute is short by design, and the score regenerates on its own. The operator reviews the verdict log periodically; sanctions that turn out to be wrong are cleared at that time, no action needed on your side beyond the fact that your account page already shows the verdict for you to see.
Every threshold and duration on this page is read live from the realm's own configuration. When the operator changes a value, this page shows the new value on the next page load.
Last reviewed: 2026-05-23. The operator updates this page whenever the data-handling practices change.