Kava said the incident on November 13 was not an unlimited token mint and was isolated to Kava Lend. In its official incident updates on X, the project also said partner exchanges had paused KAVA deposits and withdrawals as a precaution. Kava later reported that the Axelar bridge had been suspended while it prepared a full report.

The exploit appears to have been a pricing and accounting failure rather than a bug that created new KAVA. The attacker manipulated the balance of a nearly empty BUSD lending market, then used the resulting collateral valuation to borrow KAVA from the protocol.

How the accounting failure worked

Bitquery investigator Gaurav Agarwal reconstructed the exploit from onchain data, reporting that the attacker used a small BUSD loan to reduce the market’s net supply to approximately two hundred-millionths of a dollar. Over the next three minutes, the market’s BUSD index increased by roughly 243,000 times.

KAVA borrowed in the 127-loan loop, retained, moved toBNB Chain and sold.KAVA020M40M60MBorrowed in 127 loans55.2MKept by two wallets53.9MSent to BNB Chain49.2MSold on BNB Chain48.4MChart: theUnhashed · Data: bitquery.io
KAVA borrowed in the 127-loan loop, retained, moved to BNB Chain and sold. · Chart: theUnhashed · Data: bitquery.io

That multiplication transformed a relatively small deposit made by the attacker into collateral that the protocol valued at approximately 1.25 billion BUSD. The attacker then borrowed about 55.2 million KAVA, worth approximately $3.76 million before the market reacted.

The incident highlights an important distinction in decentralized finance. A protocol does not need an unlimited minting function to suffer a major loss. If its accounting system treats a tiny positive balance as economically equivalent to a healthy market, an attacker can create artificial purchasing power without changing the total supply of the borrowed asset.

KAVA price during the exploit period, including its nearly 40% 24-hour fall after the incident. · Live chart: TradingView

Kava’s code repository, Kava-Labs/kava, contains the blockchain and application code relevant to reviewing the lending market’s accounting and liquidation logic. The reported failure appears connected to the way interest and market values were calculated when supply was close to zero. The code avoided a division-by-zero error when supply was exactly zero, but did not adequately constrain the result when supply was merely extremely small.

That difference matters. A zero value can trigger a clear exception or shutdown condition. A tiny positive value can instead produce a valid mathematical result that is completely detached from economic reality. In this case, the reported index increase allowed the protocol to recognize billions of dollars in nominal collateral where almost no BUSD market liquidity existed.

Kava Lend module balance fell as KAVA was borrowedKAVA0204060Module before55.33Module after0.19KAVA borrowed55.2
Kava Lend module balance fell as KAVA was borrowed

Borrowing cap was apparently bypassed

The attacker did not rely on a single borrowing transaction. Bitquery reported 127 KAVA borrowing cycles, with a second account apparently used to bypass the protocol’s borrowing cap.

The Kava Lend module account fell from about 55.33 million KAVA to approximately 0.19 KAVA in around 90 minutes, according to the onchain reconstruction. That suggests the exploit was not limited to a test transaction or a failed attempt. Once the inflated collateral value was accepted, the attacker repeatedly extracted available KAVA from the lending pool.

PublicAML traced the preparation and movement of funds, reporting that the attacker’s preparation began at least several days before the drain. Its analysis linked activity across Kava and BNB Chain and described repeated deposit and borrowing cycles beginning around November 8.

The preparation is significant because it suggests the transaction that exposed the exploit was not an isolated act of opportunistic trading. The attacker appears to have spent time positioning accounts, testing the market and preparing a route for moving the borrowed tokens across chains.

Approximately 49.2 million KAVA was later bridged to BNB Chain through LayerZero, according to the investigations. Most of that amount was sold for roughly 3,006.5 BNB and 231,557 USDT, with the combined proceeds valued at around $2.09 million at the time described by the reports. PublicAML also said Relay blocked a USDT transfer worth approximately $612,416.39.

The cost of inactive markets

KAVA fell nearly 40% over 24 hours following the exploit, while exchange suspensions added friction for users trying to move the token. Kava’s statement framed the problem as isolated to Kava Lend, but that description does not eliminate the broader design question: how should a protocol protect users when one market can drain assets held by a shared module?

Dormant markets are often kept online for convenience. They preserve historical positions, avoid migration work and allow developers to reactivate a market if demand returns. The tradeoff is that an inactive pool may receive less monitoring than a liquid one, even though its accounting parameters remain capable of influencing borrowing power.

A safer design would treat near-zero liquidity as a distinct state rather than a normal market condition. Lending protocols could pause borrowing when supply falls below a minimum threshold, cap the maximum exchange-rate increase between blocks, impose absolute limits on collateral growth, or require governance approval before an inactive market becomes usable again. These controls would not replace code audits, but they could limit the amount available to an attacker when a calculation behaves unexpectedly.

The episode also raises questions about the meaning of an isolated-market guarantee. If a market is technically separated but its module controls a large pool of KAVA, the failure may remain local in code while becoming broad in financial impact. Users do not experience protocol architecture as a diagram. They experience it through the assets that become unavailable, the token price that falls and the bridges or exchanges that stop processing transfers.

Kava had not published an official loss figure or a complete post-mortem in the supplied material. The critical accounts from Bitquery and PublicAML provide the available reconstruction, while no opposing technical account was identified in those sources. The next test for Kava will be whether its full report explains not only how the BUSD index grew so dramatically, but also why a dormant market remained able to convert that value into borrowable KAVA.

#Kava Lend#Kava#BUSD#Bitquery#PublicAML#Axelar#LayerZero

Jessica Jones is not a person. No notebook, no deadlines, no face behind the name — just a byline this newsroom publishes under. Here is the production line underneath it, because a name beside a portrait reads like a journalist, and this one is not one.

The models. Writing: gpt-5.6-luna. Out on the live web: gpt-5.6-luna and gpt-5.6-terra. Pictures: gpt-image-1. Swap one in the newsroom and this line swaps with it — it is read off the machines, not typed here.

How a story is made

  • Research. The searching model reads around the story, pointed at primary sources — the filing, the post, the repository — rather than at somebody else's write-up of them.
  • Writing. The writing model drafts it against what was found, at Jessica Jones's usual length and in Jessica Jones's usual register.
  • The loop. A reviewer reads the draft and sends it back with notes. Then reads it again. A piece can go round several times before it leaves the building.
  • Enrichment. A quotation has to appear word for word on the page it is taken from. A chart may only use figures that appear in the source it cites. Whatever fails is dropped, and the reason is kept.
  • Fact check. A last pass hunts for claims the article makes and its sources do not.
  • A human stop. Sensitive subjects are held for a person to read before publication, and a person can kill any of it at any point.

If that sounds less like a newsroom and more like a factory: quite. It is called Press Factory.

This article was generated using AI and published automatically without human pre-publication review.

Read and checked by admin on 11/16/2026

How this article was made

The article was produced by the Grandmonts Media News Engine using automated research, drafting and verification workflows. No human editor reviewed the article before publication. Grandmonts Media remains responsible for the published content. Errors can be reported at office@grandmonts.cz.