Mentioned in This Article
1. sBOLD exploit drains about $3.43 million
About $3.43 million was taken from sBOLD, a yield vault operated by K3 Capital on top of the Liquity V2 lending protocol, on Wednesday, November 18, 2026, according to blockchain security firm Blockaid. K3 Capital said it was investigating malicious activity and has paused the vault's contracts. No cause and no final loss figure have been published.
Blockaid raised the alarm publicly with a post on X at 12:50 UTC. The firm described the sBOLD exploit as "ongoing" at that moment, and it was careful about the number it attached to it: the $3.43 million estimate covered roughly 34 transactions "so far." That wording matters, because it means the figure was a snapshot taken while funds were still leaving the vault, and the total may have risen after the alert went out.
K3 Capital answered on the same platform at 13:08 UTC, less than 20 minutes after the warning appeared. Its statement was brief. The team said it was looking into "malicious activity involving sBOLD" and added that "Contracts have been paused." The post did not put a dollar value on the losses, did not explain how the attackers got in, and did not describe what the pause means for people whose funds remain in the product.
A follow-up post from Blockaid pointed to one transaction in particular. That transaction was recorded on Ethereum (ETH) at 12:25 UTC, which places it about 25 minutes ahead of the security firm's first public warning. The on-chain record shows it contained 231 separate token transfers, an unusually dense sequence of movements for a single transaction and a sign of how many steps the attackers packed into one operation.
2. What is sBOLD?
sBOLD is a tokenized vault that accepts BOLD, the dollar-pegged stablecoin issued by Liquity V2, and puts those deposits to work in Liquity's stability pools. It follows ERC-4626, the Ethereum standard for vaults that issue shares representing a claim on a pool of underlying assets. A depositor hands over BOLD and receives vault shares in return; redeeming those shares later returns a proportional slice of whatever the vault holds.
The yield comes from the role stability pools play inside Liquity. When a loan on the protocol falls below its required collateral level and is liquidated, the stability pools supply the stablecoins needed to pay off that loan's debt. In exchange, the pools receive the collateral that had been backing the loan. Depositors in a stability pool therefore see part of their BOLD used up over time and replaced by collateral assets, which is why a vault built on those pools needs a way to value and convert what it receives.
That is where the vault's own machinery comes in. Liquity's blog describes sBOLD as a third-party project built by K3 Capital rather than a component of Liquity's core protocol. The distinction is significant. Liquity V2's contracts are immutable, meaning they cannot be changed after deployment. The vault sitting above them is a different matter: it includes parts that can be adjusted, among them the price oracle it relies on and the adapter it uses to swap tokens.
For users, the appeal of a product like sBOLD is convenience. Instead of depositing into stability pools directly and managing the collateral that liquidations hand back, a holder owns a single share token and leaves the handling of those flows to the vault. The trade-off is that the depositor takes on the risks of the vault's own code and configuration in addition to those of the lending protocol underneath.
3. How did the exploit work?
According to Blockaid, the attackers combined a flash loan with a liquidation of their own borrowing position to obtain vault shares at an artificially low cost, then cashed those shares in for BOLD that belonged to other depositors. The security firm has given an outline of the sequence, although K3 Capital itself has not confirmed any of the mechanics.
The first step, in Blockaid's account, was a flash loan denominated in wstETH, the wrapped form of staked ether issued through Lido. A flash loan is borrowed and repaid inside a single transaction; if the repayment does not happen, the whole transaction is reversed as though it never took place. Because no lasting credit risk is involved, such loans require no collateral up front, which lets an attacker command a very large amount of capital for the few moments an exploit needs.
With that capital in hand, the attackers liquidated their own Liquity V2 trove, the protocol's term for an individual borrowing position. A liquidation pushes the position's debt and its collateral into the stability pools, the same pools where sBOLD keeps its depositors' funds. In other words, the attackers used a position they controlled to trigger precisely the kind of event the vault is designed to absorb.
Blockaid said the attackers then "minted cheap sBOLD shares" and "swapped the collateral" before redeeming BOLD that was owed to other users of the vault. In a share-based vault, the price of a share depends on how the vault values everything it holds at that instant. If shares can be created for less than they are truly worth, whoever mints them ends up with a claim on assets that other depositors contributed, and redeeming those shares transfers that value out of the vault.
The security firm's follow-up post identified three addresses connected to the incident: one belonging to an exploiter, one for a helper contract used in the operation, and the address of the vault that was drained. Publishing such addresses allows exchanges, analytics companies and other protocols to watch for the stolen funds as they move.
4. An earlier audit looked at self-liquidation
Self-liquidation was not an unfamiliar idea to the people who reviewed the vault's code. ChainSecurity, a smart contract auditing firm, completed an audit of sBOLD on May 19, 2025. The summary of that review said the possibility of extracting value through self-liquidations "has been mitigated."
Whether the attack on November 18 travelled the same route the auditors had considered is not known. K3 Capital has not stated a cause, and Blockaid's description, while it mentions a self-liquidation, does not say that the weakness was the one addressed in the 2025 review. It is possible for an attack to share a general technique with a previously studied risk while exploiting a different flaw, and it is equally possible for conditions around a contract to change after an audit is finished.
That second point is relevant here because of how the vault is built. An audit examines code and settings as they exist at a particular time. The sBOLD vault contains adjustable elements, including its price oracle and its token-swap adapter, so the system in operation on the day of the exploit is not necessarily identical in every respect to the one that was reviewed. Nothing published so far indicates which component, if any, was at fault.
5. Why the line between Liquity and the vault matters
The incident centers on a product built on Liquity V2, not on the lending protocol's own contracts. Blockaid's alert concerned the sBOLD vault, and K3 Capital's response referred to activity "involving sBOLD." Liquity's public description of the vault as an independent project from K3 Capital draws the same boundary.
This layering is common across decentralized finance. A base protocol provides a narrow function, in this case borrowing against collateral and the stability pools that backstop those loans. Other teams then build products that package the base protocol's features into something simpler to hold. Each added layer brings its own contracts, its own assumptions about prices, and its own points of control.
The attack described by Blockaid appears to have used Liquity V2's ordinary features as tools. Opening a trove, being liquidated, and having debt and collateral flow into the stability pools are all standard behaviors of the protocol. What the attackers allegedly gained was a pricing advantage inside the vault while those flows were happening. The ability of K3 Capital to pause sBOLD also reflects the difference between the two layers: an immutable protocol cannot be halted by its developers, while the vault evidently could be.
6. What remains unknown
Several basic questions are still open. The $3.43 million estimate reflects losses as Blockaid measured them at 12:50 UTC, when it said the exploit was continuing. No final tally has been released by the security firm or by K3 Capital, so the true size of the loss could be larger than the first figure suggests.
The cause has not been disclosed either. K3 Capital's statement confirmed an investigation and a pause but offered no technical explanation, and Blockaid's outline describes what the attackers did without stating which specific weakness made it possible.
Depositors also have no guidance yet on their position. K3 Capital's post did not say what users of the vault can expect, whether in terms of access to remaining funds while the contracts are paused or any plan to address the losses.
7. What comes next for sBOLD depositors
The next developments are likely to come from K3 Capital's investigation. A fuller account from the team would be expected to address how the attack worked, how much was ultimately taken across all the transactions involved, and under what conditions the paused contracts might be reopened. None of those points has a published timetable.
Until then, the public record consists of Blockaid's alert and follow-up, K3 Capital's short acknowledgment, and the transactions visible on Ethereum. Together they establish that funds left the vault, that the operator has stopped the contracts, and that a flash loan and a self-liquidation were part of the method as the security firm understands it. They do not yet establish the root cause, the final cost, or what the outcome will be for people who deposited BOLD into the vault.
