Security Guides

Double Counter Discord bot data breach (about 28 million Discord IDs, 1 million emails): what leaked and what members and server admins should do

The Discord bot Double Counter says an October 4, 2026 intrusion exposed about 28 million Discord IDs, 27 million IP addresses and 1 million emails. What leaked, and what members and server admins should do.

Published 2026-10-30 Updated 2026-10-30 Last verified 2026-10-30 11 min read
On this page

For: Discord users who verified through the Double Counter bot, administrators of servers that use it, and people who run self-hosted analytics tools. This article is based on the operator's own report and does not cover attack techniques.

Developing: this article will be updated as the organisation publishes more

As of October 30, 2026

Not yet known

  • Exact number of people affected
  • Which published data copies have been removed
  • Result of the data protection authority review
  • Result of the criminal complaint

The operator's report is dated October 5 and was updated to version 2.0 on October 21. It says the incident is contained, service was restored with new credentials, and removal of published copies of the data has been requested. As of October 30 we found no later version.

What to do today

1

Members: do not act on messages that use your Discord ID, email or IP address

The leaked items let someone write a message that looks specific to you, for example "your Double Counter account is affected" with a link. The operator says it never asks for passwords, tokens, or payment by email or private message. If you receive an email about the incident, do not click the link; open the operator's website yourself (see What is phishing?).

You do not need to change your Discord password because of this incident. A second login factor on Discord is still worth having.

2

Members: re-verify only if you were affected in the verification window

The operator says that if you verified between 13:39 and 14:49 UTC on October 4 and did not receive your role, you should verify again. Otherwise no action is needed.

3

Server admins: delete the invitation posts and review your audit log

The operator says invitations to the attacker's server were posted as if from Double Counter in about 50 large servers, starting at 13:30 UTC on October 4. Delete any Double Counter message from October 4, 12:00 to 16:30 UTC, that invites people to another server, and check your audit log for Double Counter actions in that window.

4

Bot, advertiser and API customers: treat payment-related messages as fake

The operator says it is contacting affected customers of its payment products itself. Customers should be wary of fake payment or refund messages. A card brand, expiry date and last four digits together with a name and billing address are enough for a convincing fake invoice.

5

If you have your own Discord bot, rotate its token

The operator says the attacker read the bot token from a container and used it to act as the bot. This is a general reminder, not a claim about your bot: store tokens in a secrets store rather than in plain files, restrict the permissions you give the bot, and reset the token whenever you suspect exposure.

What happened (per the operator's report)

The following is based on the incident report published by Tellter SAS, the operator of Double Counter, on October 5, 2026 and updated to version 2.0 on October 21 (checked October 30). Times are UTC.

  1. October 3, 05:28

    Probing of the operator's systems began from rotating VPN addresses.
  2. October 4, 00:47

    The attacker logged into a retired server at a hosting provider, through a publicly reachable self-hosted Metabase instance on that server.
  3. October 4, 12:03 to 17:54

    Activity inside the cloud project, 5 hours 51 minutes in total. The operator says the attacker exported a database (5.0 GB, which the report says was never downloaded), read the bot token, copied about 12 GB, and changed the database administrator password.
  4. October 4, 13:30

    Invitations to the attacker's server begin to appear in about 50 large Discord servers.
  5. October 4, 17:54 and 19:19

    The last recorded action in the cloud infrastructure, and then service restored with new credentials.
  6. October 5

    Report published. The French data protection authority (CNIL) was notified.
  7. October 21

    Version 2.0 of the report. Notification emails to about 1 million people began under GDPR, and a criminal complaint was filed with the Paris public prosecutor.
~28M
Discord IDs and usernames (partly copied, treated as exposed)
~27M
IP addresses with coarse location (partly copied, treated as exposed)
~1.0M
Email addresses, de-duplicated (copied)
~50
Large servers where invitations were posted
Data affected (per the operator)
Discord IDs and usernames
About 28 million records, partly copied and treated as exposed
IP addresses and coarse location
About 27 million records, partly copied and treated as exposed
Device details
About 25 million records of hashed user-agent, city and country, used for alt-account detection. Copied
Email addresses
About 1.0 million, de-duplicated. Copied
VPN detection logs
About 15 million. The operator says these were not copied
Payment product customers
About 5,000 records of name, email, Discord ID, billing address, plan, payment history, card brand, expiry date and last four digits. Copied, partly published
Not affected
Discord passwords, which the bot never received. Full card numbers and security codes, which only the payment provider holds
Actions taken
Deleted the stolen credential, revoked administrator sessions, removed the attacker's access key, shut down the legacy server, reset the bot token, deleted ten exposed webhooks, closed the database to the internet, rotated credentials, signed staff out and turned on monitoring

Retired server, internet-reachable Metabase

↓

Forged administrator session

Two cloud credentials found on that server (admin key, saved session)

↓

Database exported, about 12 GB copied

Bot token read from a container

↓

Bot used to post invitations in about 50 large servers

How the operator describes the intrusion (October 2026 incident report)

What is known and what is not

Status as of October 30, 2026. When the operator publishes more, we will update the status column.

ItemWhat the operator saysStatus
Discord IDs and usernamesAbout 28 million; partly copiedTreated as exposed (operator)
IP addresses and coarse locationAbout 27 million; partly copiedTreated as exposed (operator)
Hashed device detailsAbout 25 millionLeak confirmed (operator)
Email addressesAbout 1.0 million, de-duplicatedLeak confirmed (operator)
VPN detection logsAbout 15 millionNot copied (operator)
Payment product customer recordsAbout 5,000; partly publishedLeak confirmed (operator)
Discord passwordsNever received by the botNot affected (operator)
Full card numbers and security codesHeld only by the payment providerNot affected (operator)
Abuse of the botInvitation posts in about 50 large servers; substantially all deletedConfirmed abuse (operator)
Card charges on a separate Tellter product23 of 40 attempted charges succeeded for US$7,331; refundedConfirmed (operator)
Entry pointPublicly reachable self-hosted Metabase on a retired server; forged administrator sessionConfirmed (operator); CVE not named
Backdoors left behindNone found in an audit of 14 cloud projectsNo backdoor found (operator)
Regulator and law enforcementCNIL notified October 5; criminal complaint filed October 21Confirmed (operator)
Removal of published copiesRequestedUnder way

How to read this: 'partly copied' is the operator's wording

The operator treats the Discord ID and IP address records as exposed even though only part was copied. The counts are approximate, in the operator's words. This site does not add the categories together, because the same person may appear in several of them.

What these items can be used for

In general, these combinations can be misused as described. This does not mean it has happened in this case.

  • Discord ID and username: lets someone message you directly or impersonate a bot or a moderator in a server you share
  • IP address and coarse location: tells someone roughly where you are and which internet provider you use; it is not a street address, but it makes a threat or a "we know where you are" message sound real
  • Email address with a Discord ID: links your Discord identity to your email, which can be used to target you outside Discord
  • Billing details of payment product customers: a name, billing address and card tail are enough for a believable refund or invoice scam

For people who run analytics tools and bots

The operator says the entry point was a self-hosted analytics tool on a server it had retired but not removed from the internet, and that the attacker then used two cloud credentials that were stored on that same server. This follows other breaches through the same kind of tool; see our article on Metabase vulnerability breaches. The operator did not name a CVE for this incident, so we do not link it to that one.

Two checks follow from the operator's own account:

  • Retired systems: list every server and tool you still pay for or still have DNS for, and delete the ones you no longer use. Check from outside your network which of your analytics, dashboard and admin tools answer on the internet.
  • Credentials that can do everything: the operator reports the attacker used a service-account key with administrator rights and a saved administrator session. Give service accounts only the permissions they need, set short lifetimes for saved sessions, and alert on a new SSH key or a first-time login to your cloud project.

For Discord bots specifically, the token is the bot's identity. Anything that can read it can act as the bot in every server where the bot has permissions, so keep the permissions you grant the bot as narrow as possible.

Sources (public record)

The facts in this article are based on the public information below. We do not speculate on undisclosed techniques or on who was responsible.

  • Tellter SAS (operator of Double Counter), "Security incident INC-2026-10-04" incident report (October 5, 2026; version 2.0, October 21, 2026) — doublecounter.gg

Update history

2026-10-30: First version, based on the operator's report version 2.0 (October 21), checked October 30. We will update this article when the operator publishes a new version.

Latest breaking reports

All breaking reports →

FAQ

QWhat leaked from Double Counter?
A

According to the operator's report, about 28 million Discord user IDs and usernames, about 27 million IP addresses with coarse geolocation, about 25 million hashed device details (hashed user-agent, city and country) and about 1.0 million de-duplicated email addresses. The operator says the first two categories were partly copied and are treated as exposed, and the other two were copied. About 5,000 customer records of its payment products were also copied.

QAm I affected? Do I need to change my Discord password?
A

If you verified through Double Counter in a server, your Discord ID may be among the records. The operator says Discord passwords were never received by the bot, so there is nothing to reset, and members need no change to their Discord account. If you verified between 13:39 and 14:49 UTC on October 4 and did not receive your role, the operator says to verify again.

QWhat happened with the invitation messages in servers?
A

The operator says the attacker used the bot's token to post invitations to another Discord server, which appeared to come from Double Counter, in about 50 large servers from 13:30 UTC on October 4. Substantially all have been deleted. Server administrators should delete any Double Counter invitation message from October 4 between 12:00 and 16:30 UTC and check their audit log. Members should not join servers advertised in unexpected bot messages.

QWere card details exposed?
A

The operator says full card numbers and security codes are not affected because only the payment provider holds them. About 5,000 customer records of its payment products, which include name, email, Discord ID, billing address, plan, payment history, card brand, expiry date and last four digits, were copied and partly published. Separately, the attacker used a stolen payment key on a different Tellter product to charge 11 customers' saved cards; the operator says the charges were refunded.

QWhat caused it?
A

The operator says the attacker logged into a retired server at a hosting provider through a publicly reachable, self-hosted Metabase instance. A vulnerability allowed a forged administrator session. The report says the retired server held two credentials for the operator's cloud, a service-account key with administrator rights and an administrator's saved command-line session, and the attacker used them. The operator names neither a CVE nor the specific vulnerability, so we do not link it to any specific Metabase flaw.

QWhat if someone contacts me claiming to be Double Counter?
A

The operator says it never asks for passwords, tokens, or payment by email or private message. If you receive an email about the incident, do not click links in it; go to the operator's website yourself. Advertisers, API users and customers of its payment products should expect phishing, including fake payment or refund messages.