The loss was small, and that is why it matters. A protocol lost almost all of its operating money not through a flaw in its own trading rules, but through a login system it trusted to guard the keys that move that money. The danger sat outside the protocol, in shared infrastructure, and it reached a treasury through an ordinary internet-connected wallet used to pay ordinary bills.
- Frontend & Infrastructure6,51574%
- Key Compromise7238.2%
- Bridge & Cross-Chain7028.0%
- Access Control3714.2%
- Token & Share Accounting2582.9%
- Other2342.7%
DefiLlama's public database of hacks, dated 19 November 2026, records that incident alongside another case from the same day, and places both inside a year in which reported thefts have grown sharply. The database lists the day, project, networks, estimated amount, attack type and any funds returned for each incident that projects, auditors or investigators reported publicly.
A treasury emptied through a login
The first case concerned CALC, a DeFi protocol operating on Ethereum and Hyperliquid L1, which lost about $26,900. DefiLlama puts it in the Frontend and Infrastructure category, with a key leaked through infrastructure, which covers failures in surrounding web and login systems rather than a mistake in the protocol's trading rules. CALC said the money it keeps for running costs was nearly wiped out after attackers reached a wallet opened through Web3Auth and Keplr social login. CALC reported that a weakness in Web3Auth's older infrastructure exposed signing credentials that authorise withdrawals, and money left the affected accounts without permission.
To pay for off-chain infrastructure, CALC used a Keplr hot wallet, a wallet connected to the internet. That everyday use is the point: a wallet kept online for routine expenses can also move a treasury, so a leak in a login provider can empty it in one move. CALC reported losing 99.9888% of its treasury, leaving only a sliver behind. A reader who uses such a service can ask a simple question first: does the provider keep operating money in an internet-connected wallet, and who else holds the keys that can move it?
Fluidkey, the other case that day, looked different. It runs a self-custodial privacy wallet, meaning users keep their own keys and the service is built to keep transactions private. Fluidkey reported unauthorised transactions affecting its users, a flaw in protocol logic, meaning a mistake in the wallet's own rules rather than a leak elsewhere. The team shut its website, application and services at 18:05 UTC, and afterwards reported no further such transfers. It is investigating with SEAL 911 and tracing the money, while the attack method and the total loss remain undisclosed and no amount has yet been entered in the database.
The month that changed the year's total
Between 1 January and the database date, DefiLlama counted 362 incidents, with about $8.80 billion taken where an amount is known. The same part of 2025 brought 126 incidents and about $2.64 billion. Amounts in the database are estimates made at the time and can be revised, and an incident without a known amount is counted but adds nothing to the totals, so the comparison rests on reported cases rather than a complete census of every loss.
| Project | Network | Taken | Type of attack |
|---|---|---|---|
| CALC | Ethereum | $26,900 | Frontend & Infrastructure |
| Fluidkey | not recorded | not recorded | Protocol Logic |
In the last thirty days, the database counted 53 incidents and about $6.46 billion, against 31 incidents and about $515.36 million in the thirty days before them. The jump was dominated by one case, Binance CEX on 29 October, at about $6.36 billion, also classified as Frontend and Infrastructure. That dominance invites a sceptical reading: a single giant theft can make a month look extraordinary. The count answers part of that doubt, because more incidents arrived in the later period as well as more money.
Frontend failures now lead the losses
For the year so far, Frontend and Infrastructure losses stand at about $6.51 billion, far ahead of Key Compromise at about $723.03 million. A key compromise, in plain words, means an attacker obtained a private key, the secret that controls a wallet. The ranking matters because it reverses the usual picture of crypto theft as a hunt for clever flaws in trading code. Here the largest losses came through the doors around the code: websites, logins and the infrastructure that holds the keys that authorise withdrawals.
Recovery, by contrast, remains tiny. Funds returned this year where the database records them total about $8.25 million. That gap is the practical meaning of the year's figures for anyone whose money sits in a protocol or wallet: prevention, and the choice of where keys live, matter more than any hope of getting stolen funds back after the event.
What a user can check before trusting a wallet
First, ask how a service reaches its treasury: a hot wallet used for routine bills is a single point of failure if its login provider is breached. Second, ask whether signing depends on a third-party login at all, and what happens if that provider's older systems are exposed. Third, ask whether operating money is separated from user funds, so that a leak in one place cannot drain the other. None of these questions requires technical skill, only a willingness to treat a login as part of the security of the money itself.
Fluidkey's disclosure will be the next test, and the next monthly reading will show whether infrastructure failures keep leading the losses or whether the Binance case was a solitary spike. Until then the lesson of a small November loss stands: in crypto, the weakest point is often not the clever contract at the centre, but the familiar login at the edge, and the money that depends on it.
Source: DefiLlama · Date:



