On 2 November 2026 BTCPay Server said its investigation had identified an attack vector against instances running version 2.4.1, released on 23 July 2026, or older. It said it had no credible reports of hot-wallet drains on the latest release, 2.4.7. The post followed a warning late on 1 November, from the project’s X account, that some operators had reported wallets being drained, and that servers should be taken offline while that was checked. BTCPay has not published a count of servers, a total taken, or a description of how the access was obtained. This page will not invent one.
BTCPay’s blog the same day is the fuller account, and it corrects the narrower reading of the morning post. The reports were of bitcoin and litecoin taken from hot wallets. The attacks targeted version 2.4.1 or earlier. The project says the attackers used a different method to exploit the vulnerability that version 2.4.2, released on 7 August 2026, had already fixed. That is the August flaw, reached another way. It is not a statement that 2.4.7 was drained. Servers already running 2.4.2 or later when the 1 November warning went out have, BTCPay says, shown no impact from these attacks. Founder Nicolas Dorier had said the attacker logged in as an admin. The blog still does not say how that login was gained. In August the project had withheld the mechanics while people were updating. This page will not describe the method, the August one or this one.
What happened
BTCPay Server is software a merchant or a person runs themselves, so they can take bitcoin payments without handing the coins to a payment company. In a hot-wallet setup the server holds the keys and can send. That is the product. It is also why an admin login is enough to spend what sits there. The coins are not in a vault that waits for a second person. They are on the machine. See How crypto wallets actually work if a login and a key still feel like the same object.
The blog’s clock is in UTC. Version 2.4.2 shipped on 7 August. A drain described in the incident reports occurred at 13:46 on 1 November. The first email reporting a theft arrived at 20:56. At 22:25 the project warned on X and asked operators to shut servers down. Before that post it had asked hosting companies, including LunaNode, to take instances offline. BTCPay has not established when the exploitation began. A host pulling the plug is that host’s switch. It is not a refund. The blog also lists destination addresses and says some activity at them predates the reports received so far, so those balances are not a loss total. This page will not repeat the addresses.
Dorier said he runs 2.4.7 with a hot wallet himself, and then said not to put more on a hot wallet than you are ready to lose. “We can’t guarantee it won’t happen again.” Current software is not a promise that the next report will spare it. A hot wallet is a float you can spend from a server. The amount you leave there is the amount an admin session can move.
Why it matters
If you run BTCPay, you are the operator. The balance on that server is not an exchange credit and not a claim on BTCPay the project. The project does not custody the coins. Your server does, for as long as the keys live on it. A customer who already paid you has paid you. If that payment landed in the hot wallet and the hot wallet was swept, the loss is the operator’s. A buyer who holds their own keys, and who only sent a payment, was not the wallet that was drained.
Version 2.4.7 having no credible drain reports is a status from the morning post. The blog is firmer about the line that mattered: no impact seen on servers that were already on 2.4.2 or later when the warning went out. Neither sentence is a certificate for a server you installed once and never opened again. If the server was on 2.4.1 or older, including one that was updated only after the warning, BTCPay says a version bump on that same machine is not enough. Access someone may already have is not removed by the new version. The project wants a fresh server and a new wallet, even when the balance looks untouched. Keys that were never on the server, including a hardware wallet, were not the wallets that were drained. The rebuild advice still applies to the old machine. Do that from BTCPay’s own blog. A message that offers to do it, and asks for a recovery phrase, is not the blog. Reports go to security@btcpayserver.org, and the same post says never to send a seed, a private key, or a password. See How do you keep crypto safe?.
What happens next
A loss figure, if the project publishes one. The addresses it listed are not that figure. More reports from operators on old versions. The practical test is the version the server was running on 1 November, and whether that machine is still the one on the network. Watch BTCPay’s own account and its blog. We will not treat “no impact on 2.4.2 and later” as “a hot wallet cannot be drained”, and we will not treat a shutdown by a host as the coins coming back.
Not a recommendation to install, keep, or abandon BTCPay. Not a finding that every self-hosted payment server was hit. Not a description of the access. The blog of 2 November named the vulnerability’s vintage and said the method was a different one. It did not name the method.






