Mon. Nov 16th, 2026

Luxor confirms Bitcoin mining hijack as Kava traces 53.9 million stolen tokens

ByJohan Shamshad

November 16, 2026 #Kava
HackHack

Bitcoin mining infrastructure provider Luxor confirmed that attackers redirected customers’ computing power through its Commander software, while mining-pool records subsequently showed 0.324 BTC paid to an address associated with the incident. Separately, a new blockchain investigation traced proceeds from the Kava Lend exploit across multiple networks, identifying blocked stablecoins, unidentified receiving services and transactions dating back more than a week before the attack. Kava has since confirmed approximately 53.9 million tokens were stolen and announced plans to permanently disable new lending.

Luxor confirms mining software compromise affecting customer accounts

Luxor Technology disclosed on November 15 that attackers had compromised its Commander fleet-management software, temporarily redirecting Bitcoin mining hashrate from just over a dozen customer accounts toward another mining pool.

The company’s internal monitoring detected unauthorized changes to mining-pool configurations during the night of November 14–15. Its engineering team patched Commander within two hours of identifying the problem and began restoring affected machines to their original settings.

Luxor said no customer data was leaked or extracted, and that the compromise was isolated to Commander. Its Bitcoin mining pool and LuxOS firmware were unaffected, according to the company.

The machines were still doing the work, just the rewards were going somewhere else.

Luxor promised to compensate every affected customer for the diverted hashrate as though their equipment had continued mining normally. The company has not disclosed the total reimbursement amount, how many individual machines were affected or whether compensation payments have already been completed.

The first public warning came from mining infrastructure operator CoreX Hosting, which reported suspicious activity involving machines redirected to the OCEAN mining pool. CoreX identified Bitcoin address bc1qphukp9rjfc4rcej7erpm9ae20xmtspw8pcddrr as the destination for the diverted mining rewards.

Early reports claimed approximately 64,000 miners representing 13 exahashes per second had been redirected. Luxor has not confirmed either figure, reporting only the number of affected customer accounts.

Luxor CEO Nick Hansen described the breach as involving a sophisticated AI-enabled attack vector. His statement did not include technical evidence demonstrating how artificial intelligence was used or how the attackers obtained access to Commander.

The company has promised a full technical post-mortem, which should clarify the initial access mechanism, the duration of unauthorized control and the process used to change mining configurations.

OCEAN records show four Bitcoin payouts to flagged address

Mining-pool records reviewed on November 16 show that the Bitcoin address identified by CoreX received four OCEAN payouts totaling 0.32445584 BTC, overturning early reports that the suspected attackers had earned no mining rewards.

The payments were recorded at 17:21 and 22:54 UTC on November 15, followed by another two at 06:25 and 08:18 UTC on November 16. Each payment amounted to approximately 0.081 BTC.

Earlier coverage had identified only two or three payouts. OCEAN’s publicly available records subsequently showed a fourth, bringing the total recorded earnings to 0.32445584 BTC.

Then more payments arrived, even after the attack was reportedly over.

The timing reflects OCEAN’s reward-distribution mechanism, under which previously submitted mining shares can continue receiving payouts when the pool discovers additional Bitcoin blocks. A payment arriving after Luxor patched its software therefore does not independently establish that unauthorized redirection was still taking place.

OCEAN’s dashboard also showed thousands of recorded worker identifiers associated with the flagged address, although those identifiers are not necessarily individual physical mining machines. The records cannot independently establish Luxor’s total affected machine count.

Nor do the four payments establish the attacker’s final profit. The ownership of the address has not been conclusively verified, and Luxor has not established that every recorded payout resulted from compromised customer equipment.

The latest records do, however, provide a measurable financial trail for an incident that initially appeared to have produced no confirmed mining proceeds.

Kava confirms 53.9 million tokens stolen through lending exploit

In a separate security development, Kava Chain confirmed on November 16 that approximately 53.9 million KAVA tokens had been stolen from its HARD Lend lending application during the November 14 exploit.

The announcement followed earlier findings of a 55.2 million KAVA accounting discrepancy in the lending module. The larger figure reflected the gross borrowing activity and recorded deposit shortfall, while Kava’s new statement identified approximately 53.9 million tokens taken from the application.

Kava said the attackers exploited a lending-accounting vulnerability that artificially inflated the value of BUSD collateral, allowing repeated borrowing and liquidation transactions to extract existing KAVA.

Independent transaction analysis identified 127 borrowing operations and estimated that the BUSD market’s interest index had increased approximately 243,000-fold because of calculations involving an extremely small effective supply.

A dormant BUSD market, and still enough of an opening for the attacker to use.

The stablecoin had already ceased new issuance after the New York Department of Financial Services ordered Paxos to stop minting BUSD in February 2023. The token nevertheless remained supported by Kava Lend’s older lending infrastructure.

Kava’s latest statement confirmed that the exploit did not create new native KAVA tokens or compromise the underlying blockchain. The team said most individual user deposits were unaffected, although it has not published a detailed reconciliation of remaining lending obligations.

The vulnerability has now been blocked, according to Kava, and the project is accelerating the retirement of HARD Lend. It plans to submit a governance proposal permanently disabling new lending and borrowing, subject to a seven-day vote. Existing user positions will be handled through a seperate wind-down process.

South Korean exchanges Upbit, Bithumb and Coinone previously suspended KAVA transfers as the exploit unfolded, while Kava paused its Axelar bridge as a precaution.

Investigators trace Kava proceeds across multiple blockchains

A November 15 investigation by blockchain analytics service PublicAML reconstructed how the exploiter prepared the attack and distributed the resulting funds across several cryptocurrency networks.

The investigation linked Kava account kava1xltn9y20smxxav0gwsv7xpr2ww0szyzzga76lv to BNB Chain address 0x37D732914F86Cc6eb1E87419e3046A739F011042, identifying them as representations of the same underlying account identifier.

According to PublicAML, the BNB Chain wallet began preparing on November 6, receiving 13.42 BNB, acquiring approximately 8,980 BUSD through PancakeSwap and bridging the stablecoin to Kava. Additional transfers followed on November 8 and 9.

Investigators also identified a small November 6 LayerZero transfer that appeared to test the cross-chain route subsequently used during the exploit.

Between 10:02 and 12:05 UTC on November 14, approximately 49.21 million KAVA moved from Kava to BNB Chain through LayerZero.

The receiving wallet subsequently executed sales through MetaMask Swaps, CoW Protocol and Decent. PublicAML traced approximately 41.6 million KAVA through MetaMask’s swap router, 5.5 million through CoW Protocol and nearly 1.99 million through Decent.

The sales generated approximately $2.09 million in BNB and USDT, based on transaction-level estimates. That amount represents realised trading proceeds rather than the pre-attack market value of the tokens extracted from the lending protocol.

A lot of tokens moved, and the money trail split up rather quickly.

One transaction route involved 1,000 BNB converted into 612,416.39 USDT. A subsequent transfer request through Relay.link was rejected with the status BLOCKED_WALLET.

PublicAML’s examination of the Relay request found no completed payout or refund recorded as of November 15. The rejected transfer therefore represents funds prevented from completing that particular route, not an independently confirmed recovery for Kava depositors.

Another 750 BNB moved through two intermediary wallets and LI.FI routes, resulting in approximately 450,775 USDC reaching Arbitrum. PublicAML reported that the receving wallets subsequently forwarded the stablecoins, but it did not establish their ultimate destination.

Investigators also traced 1,082.58 BNB into four short-lived deposit contracts associated with a single unidentified service. Each contract was reportedly created and swept within a minute, with the funds consolidated through shared infrastructure.

The service operator remains unidentified. Transaction similarities may establish a connection between the deposit contracts without identifying who ultimately controlled the assets.

PublicAML also identified additional exits involving 500 BNB routed toward Bridgers and 27,825.46 USDT passing through Mayan-related infrastructure. The final destinations of those transfers remain unresolved.

Kava confirmed on November 16 that a portion of the stolen funds had been frozen and that ecosystem contributors, exchanges and law enforcement were cooperating on further tracing and recovery. It did not specify the amount secured or identify every service involved.

And the blocked funds are still not funds returned to depositors.

The recovery challenge has parallels with the Tectonic exploit on Cronos, where funds transferred to Ethereum remained outside the scope of an emergency rollback and were subsequently traced into Tornado Cash. The incidents involved different vulnerabilities and are not connected, but both illustrate how assets that cross into other networks can complicate recovery efforts.

For Kava, the latest investigation narrows the uncertainty surrounding how the stolen tokens were converted and distributed. It does not identify the attacker or establish the final ownership of the receiving service. The team’s newly announced safeguards prevent repetition of the identified exploit, but the outcome of the proposed lending shutdown and the amount ultimately recoverable remain open.

Luxor’s incident presents a different outstanding question: how attackers obtained sufficient control over mining-management software to redirect customers’ revenue without taking possession of their machines. The company’s promised post-mortem and Kava’s continuing recovery disclosures will determine how much of each incident can ultimately be accounted for.

Editor-in-Chief at  | More Posts

Johan Shamshad is Editor-in-Chief at Dave Finances, covering financial markets, digital assets, trading platforms, brokers, fintech, payments and regulation. Since 2023, he has also worked as an editor at Gulf News, the UAE English-language newspaper published by Al Nisr Publishing LLC. His work focuses on market-moving developments, company disclosures, regulatory changes and trading infrastructure, with an emphasis on primary-source verification and practical context for traders and investors.

Disclaimer: This content is for informational purposes only and does not constitute financial, investment, legal, or trading advice. Trading and investing involve risk. Read our full Disclaimer.