Safeguard: doxxing, IP grabbers, threats and the watchlist
Last updated: 14 October 2026
Why it exists
A group from another server found out where one of our friends lives and sent the police to their door with a false report ("swatting"). Liora can't protect anyone outside Discord. But that kind of harm nearly always starts inside it: someone's address or phone number posted in a chat, a disguised link that records your IP address, a threat that nobody took seriously. Safeguard is there for those moments, in every server Liora lives in.
What she does
- Personal details are taken down. Phone numbers, home addresses, map locations, ID and card numbers, and IP addresses posted about a person are deleted straight away, and the server's staff are told (with the details hidden). A game server's IP, a business on a map, or a famous address is left alone. If the details look like your own, she just sends you a kind word about keeping them private. If they're someone else's (a "dox"), the person who posted them also gets a 60-minute timeout while staff look, and the server's owner is told. She reads forwarded messages, posts made through webhooks (like PluralKit), small text files, and edits to old messages too, so none of those is a way round it.
- IP grabbers are taken down. Links from Grabify, IPLogger and the disguises they use (fake picture, game and YouTube links) record the address of anyone who clicks them. They're deleted, with a 10-minute timeout, like scam links.
- Threats are seen, not hidden. "I'll get you swatted", "I know where you live", "I'm leaking her address": Liora leaves the message up so staff can read it in context, keeps a locked copy in case it's deleted, and tells staff and the owner right away. She doesn't punish anyone for words. People make that call. Everyday rough talk ("i'll kill you lol") only gets a quiet heads-up in the staff log, and no copy is kept.
- The watchlist. Bliss and Mother keep a short list of accounts and servers known to have put people in danger. When a listed account joins a server Liora is in, that server's staff are told at once. Each server chooses what else happens: nothing more (the default), a 24-hour timeout while they look, or a kick. Invites to a listed server are taken down, and if someone adds Liora herself to a listed server, she tells Bliss and Mother and leaves. Liora never bans anyone for being on the list. A ban is always a person's decision.
!securitycheckshows a server's staff the doors left open (moderators without two-factor login, @everyone allowed to ping everyone, bots with Administrator, Liora's role too low to help) and how to close each one.
Staff are never timed out, and a timeout staff already gave someone is never shortened. Each part can be switched off for a server with !safeguard off, which needs the server owner's OK, like any loosening of protection. Some places do need addresses (a meet-up channel, say): staff can let personal details be posted in one channel with !safeguard exempt #channel, again with the owner's OK.
The evidence vault
Reporting a threat to Discord or the police needs the exact words, who wrote them and when, and people delete their messages once they know they were seen. So whenever Safeguard steps in, it keeps a locked copy: the message text, the author's user ID and name, the server and channel, when it was posted, the message link, attachment names and links, and what Liora did.
- It's encrypted, and only sent by DM, never shown in a channel.
- The server's owner, Bliss and Mother can see the full text. The server's admins see it with personal details hidden, so the vault can't be used to collect anyone's address. Apart from Bliss and Mother, nobody can delete a case about something they posted themselves.
!evidence exportwrites a case out as a file ready to attach to a report to Discord Trust & Safety (dis.gd/request) or to give the police.
What she keeps
| What | Why | How long |
|---|---|---|
| Evidence cases (the message, its author, where and when, what was done; at most a few per person per minute) | So a threat or a dox can be reported even after it's deleted | 90 days; the server's owner can keep a case longer for a report that takes a while, but never more than a year after it was saved. When Liora has left a server for good, Bliss's clean-up tool removes its cases with the rest of its data |
| The watchlist: listed accounts' user IDs, the name they had, why they're listed, who listed them and when; listed servers' IDs and names | So every server Liora is in knows who to look out for | Until Bliss or Mother remove them. Every entry comes up for a second look after 180 days |
| Which invite links lead to which server | To spot invites to a listed server | In memory only, 6 hours |
Personal details that Safeguard takes down are also removed from Liora's 7-day message cache and hidden in her own chat memory, and they're hidden in the logs, alerts and reports she posts for staff (edits, deletions, spam, profanity and ping alerts, reported messages and her own reports), along with what the lines of a dox list say (name, school, parents...), so a log channel never becomes a second copy of a dox. Liora's encrypted nightly backups (kept 14 days) include the vault, like everything else.
Your control
/privacy mydatatells you how many evidence cases are about you (something you posted, or you joining a server while on the watchlist), and whether you're on the watchlist./privacy forgetmedoesn't remove either. They're safety records, like a warning.- If you think you're on the watchlist by mistake, talk to Bliss (@thenamesbliss). If a case is about you and you think it was a mistake, talk to that server's staff.
The lines
Doxxing someone, sending IP grabbers, or threatening to swat, dox or turn up at someone's home can get you added to the watchlist, and Bliss can block you from using Liora anywhere. It may also be reported to Discord and to the police. Swatting is a crime in many countries, and it has cost people their lives.