LayerZero has attributed the responsibility for a $290 million exploit involving Kelp DAO to Kelp’s own security configuration. The company stated that Kelp employed a single-verifier setup, which LayerZero had previously advised against.
The attackers exploited an innovative vector targeting the infrastructure layer rather than any code within the protocol itself. With preliminary confidence, LayerZero linked the attack to North Korea’s Lazarus Group and its TraderTraitor subunit.
The breach involved compromising two of the remote procedure call (RPC) nodes that LayerZero relied on for verifying cross-chain transactions. These nodes are essential servers enabling software interaction with blockchain data. By substituting the binary software on these nodes with malicious versions, the attackers deceived LayerZero’s verifier into acknowledging a fraudulent transaction while maintaining accurate reporting to other systems.
This deception was specifically crafted to evade detection by LayerZero’s monitoring infrastructure, which queries RPCs from various IP addresses. Additionally, the attackers executed a distributed denial-of-service (DDoS) attack on uncompromised external RPC nodes to force reliance on the compromised ones. Traffic logs provided by LayerZero indicate that this DDoS occurred between 10:20 a.m. and 11:40 a.m. Pacific Time on Saturday.
Following the failover, the corrupted nodes falsely informed the verifier of a legitimate cross-chain message, prompting Kelp’s bridge to release 116,500 rsETH to the attackers. Subsequently, the malicious software self-destructed, erasing binaries and local logs.
This attack succeeded because Kelp operated with a 1-of-1 verifier configuration, making LayerZero Labs the sole verifier for messages related to the rsETH bridge. LayerZero had recommended a multi-verifier setup to ensure redundancy and consensus among independent verifiers, which could have prevented such an exploit by not allowing one poisoned data feed to validate a message.
LayerZero confirmed that no other applications on its protocol were affected. The company has resumed verifier operations but will cease signing messages for any application using a 1-of-1 configuration, necessitating a shift away from single-verifier setups across the protocol.
The nature of this architectural distinction is crucial for assessing LayerZero’s risk in DeFi pricing. A bug at the protocol level would have indicated potential risks across all OFT tokens on every chain, but the actual event—a targeted infrastructure attack exploiting Kelp’s security choices—demonstrates that the protocol functioned as intended.
Kelp has yet to respond publicly to LayerZero’s assessment or explain its decision to maintain a 1-of-1 verifier setup despite explicit recommendations against it.
The Lazarus Group, previously linked to the Drift Protocol exploit on April 1, is now associated with the Kelp incident on April 18. This marks over $575 million drained from DeFi in just 18 days through two distinct attack methods: social engineering governance signers at Drift and compromising infrastructure RPCs at Kelp.
The adaptability of Lazarus Group’s strategies is outpacing the defensive measures implemented by DeFi protocols.