Kava Lend Exploit Used a Dormant BUSD Market to Borrow 55.2 Million KAVA

Daily Feed
Kava Lend Exploit Used a Dormant BUSD Market to Borrow 55.2 Million KAVA

Kava Lend exploit: How a dormant BUSD market enabled 55.2 million KAVA in borrowing

A mostly inactive BUSD market on Kava Lend became the entry point for a major lending-pool drain. Bitquery says an accounting flaw turned a tiny deposit into an enormous recorded collateral balance, helping two wallets borrow more than 55 million KAVA.

How the BUSD market became exploitable

Bitquery dates the attack to Nov. 14, 2026. The market was already deeply imbalanced. At 04:00 UTC, Kava Lend’s BUSD pool owed lenders 440, 145 BUSD, had 5, 084 BUSD out on loan, held 0.003 BUSD in cash, and had 7, 618 BUSD in reserves, according to Bitquery.

Those figures matter because, according to Bitquery’s analysis of the protocol’s accounting, the deposit-growth calculation used a denominator based on cash plus loans minus reserves. With the reported starting balances, that denominator was negative: reserves exceeded the combined cash and loans.

Bitquery says the code accounted for a denominator of exactly zero, but not one close to zero. After the main wallet repaid the market’s 37 open BUSD loans, it made a series of deposits, loans, and withdrawals, bringing the denominator down to just two base units, the smallest BUSD amount the chain could count. The supplied transaction analysis does not provide enough detail to independently reconstruct every change in the denominator.

According to Bitquery, the supply index, the factor used to track deposit value, then jumped from 2.76 to 671, 187. That is roughly a 243, 000-fold increase. The main wallet’s recorded deposit claim grew with it, eventually reaching 1.26 billion BUSD.

That was not 1.26 billion redeemable BUSD sitting in the pool. It was an accounting balance. Bitquery says the wallet withdrew the actual 7, 618 BUSD in reserves while keeping the inflated claim in the protocol’s records.

Borrow, liquidate, repeat

Bitquery links two wallets through the activity: a main wallet that manipulated and moved BUSD, and a helper wallet that used BUSD deposits as collateral to borrow KAVA. The investigation counted 127 borrowing rounds between 04:04 and 05:42 UTC, over 98 minutes. This transaction pattern links the wallets’ roles, but does not establish who controlled them. Bitquery’s on-chain investigation of the Kava hack includes related transaction analysis.

The repeated borrowing relied on how liquidation worked in this case. Bitquery says the main wallet moved the real BUSD before the helper wallet’s loan was liquidated. Liquidators then found only zero, one, or two base units of BUSD to auction.

The loan was still formally closed through liquidation. According to Bitquery, that freed up capacity under the KAVA market’s borrowing cap, letting the helper wallet borrow again. The liquidation did not recover enough collateral to cover the debt. It cleared the loan from the protocol’s accounting, leaving the underlying shortfall in place.

Bitquery reports that the helper wallet borrowed 55, 205, 051 KAVA. The two wallets kept 53.9 million KAVA after repayments, including about 1.3 million KAVA repaid for other borrowers. These figures describe different stages: the total borrowed and the amount reportedly left with the wallets. Bitquery says the KAVA pool fell from 55.3 million KAVA before the borrowing loop to 519 KAVA afterward, prompting a shutdown of Kava’s HARD Lend market.

Not a mint, but still a pool drain

Kava’s response, quoted by Bitquery, described the incident as isolated to Kava Lend and “not an infinite mint.” The distinction matters: the reported mechanism took KAVA from the lending pool rather than creating new KAVA tokens. The incident was also described as a Kava Lend pool drain.

But the absence of new tokens does not mean there was no loss. When assets deposited by users are removed from a lending pool, those users can still face a shortfall. “Not a mint” describes how the KAVA entered circulation. It does not tell us who ultimately bears the damage.

Bitquery says Kava paused a bridge as a precaution, but the available details do not identify which bridge was paused or explain how that response relates to claims about the route used to move the attacker’s funds. They also do not establish how much, if anything, was recovered or returned to users. Those outcomes cannot be inferred from the pool-flow figures alone.

Why a dormant BUSD market remained a live risk

In February 2023, Paxos announced that it would stop issuing new BUSD effective Feb. 21, following direction from the New York Department of Financial Services. Paxos said Paxos-issued BUSD was backed one-to-one by segregated, dollar-denominated reserves.

That statement does not show that every token labelled BUSD on every chain had the same issuance or redemption arrangements. Bitquery says the BUSD involved in Kava Lend had arrived years earlier from Binance Chain, now known as BNB Beacon Chain. The available information does not establish whether the tokens in the Kava market were Paxos-issued BUSD or a bridged representation, or whether they could be redeemed directly through Paxos.

Stopping new issuance did not automatically shut down existing markets. Kava Lend’s BUSD market stayed open, leaving its contracts and accounting rules active after BUSD had fallen out of regular use.

Bitquery also found small BUSD deposits, loans, and withdrawals by the main wallet between Nov. 8 and 10. That activity fits with testing the market before the larger borrowing loop, but it does not prove intent or identify who controlled the wallets.

What protocols can learn

The main lesson is to test edge cases. A formula that works in a liquid market can become dangerous when reserves, loans, and cash fall into an unusual relationship. Security testing needs to cover negative and near-zero values, not just the conditions a protocol expects on an ordinary day.

Possible safeguards include pausing new borrowing when market liquidity falls below a set threshold, limiting how quickly deposit indexes can grow, and monitoring older markets that still accept collateral. These measures can reduce exposure, but they cannot replace sound accounting or careful review of liquidation logic.

Liquidation also needs to distinguish between closing a position and recovering its debt. If a protocol restores borrowing capacity after seizing almost no collateral, the same shortfall may be exploited again.

Not every quiet market is a ticking time bomb. But low activity is no security control, and an old market can remain exploitable as long as its contracts still run.

Key questions and answers

  • What was exploited?

    Kava Lend’s BUSD-market accounting. Bitquery says a near-zero denominator caused the supply index and a wallet’s recorded deposit claim to surge.

  • Was 1.26 billion BUSD available to withdraw?

    No. Bitquery describes that as an inflated accounting claim, not real BUSD held in the pool.

  • How much KAVA was involved?

    Bitquery reports 55, 205, 051 KAVA borrowed and 53.9 million retained by the two wallets after repayments.

  • Why did liquidation not stop the loop?

    Bitquery says almost no BUSD remained to seize, yet liquidation closed the loan and reopened borrowing capacity under the KAVA market’s cap.

  • Did the exploit mint new KAVA?

    Kava said it did not. The reported KAVA came from the lending pool, so the absence of a mint does not mean the pool avoided losses.

  • Were funds recovered?

    The available accounts do not establish whether funds were recovered, frozen, or returned to users.

Share this article

Powered by ADBYTES

Advertise smarter.

Adbytes.Media is a transparent advertising network where advertisers reach real audiences and publishers, affiliates & everyday members earn ADBYTES tokens. Join the community and start earning today.

Back to Blog