Security vulnerabilities that shake the entire DeFi ecosystem

Author: thedefinvestor Translation: Shan Ouba, Jinse Finance

Full Analysis of the rsETH Attack Incident

Last week, Kelp DAO experienced one of the largest DeFi fund thefts in recent times.

The hacker used forged cross-chain messages to breach the LayerZero-supported Kelp DAO rsETH cross-chain bridge, creating 116.5k rsETH out of thin air, worth approximately $290 million.

The attack itself was already severe, but the deep integration of rsETH in the DeFi ecosystem further amplified the disaster’s impact. For example, rsETH was once listed as a compliant collateral asset on Aave.

After the hacker minted rsETH out of thin air, they immediately used it as collateral to borrow ETH on Aave, directly causing Aave to incur over $116.5k in bad debt.

Not only Aave, but the scope of this incident is very broad: protocols such as Compound, Lido’s EarnETH vault, some Morpho lending pools, Hyperithm’s mHyperETH product, Superform’s SuperWETH vault, and others, all affected to varying degrees due to holding or interfacing with rsETH.

Who is ultimately responsible for the incident?

Compared to past attack events like Drift, the responsibility for this incident is more complex to define.

The breach was through LayerZero’s rsETH cross-chain bridge, not a vulnerability in Kelp DAO’s own smart contracts. Currently, parties are passing the buck: LayerZero blames Kelp DAO, while Kelp DAO believes the full responsibility lies with LayerZero.

Objectively, the core facts are as follows:

1. The hacker compromised two RPC service providers relied upon by LayerZero’s Distributed Verification Nodes (DVN), enabling data tampering and malicious minting;

2. Kelp DAO’s rsETH cross-chain bridge uses a single-signature verification mechanism (1/1 DVN), relying on only one verification node to approve transactions, making forged transactions easily approved;

3. LayerZero accuses Kelp DAO of choosing a low-security single-node verification, but LayerZero itself tacitly permits and allows all projects to use the minimal 1/1 verification mode;

4. Before the attack, 47% of decentralized applications connected to LayerZero’s cross-chain infrastructure were using the 1/1 DVN configuration, not just Kelp DAO.

Disregarding complex technical details, it’s clear: LayerZero should bear the main responsibility and face up to its design flaws.

Kelp DAO’s oversight was oversimplifying security by relying solely on a single verification node; if multi-signature, multi-node verification had been used, this attack could have been prevented. But ultimately, if LayerZero’s RPC nodes had not been compromised, the theft would never have occurred.

Follow-up developments and industry responses

Fortunately, nearly one-third of the stolen assets have been frozen and recovered by Arbitrum, and the authorities have locked down the hacker’s related funds.

From a decentralization perspective, on-chain project teams actively freezing assets is controversial. But in practical terms, given that layer-2 networks cannot be fully decentralized, taking proactive loss mitigation measures to protect user assets is more meaningful than just talking about ideals.

Meanwhile, Aave is evaluating multiple solutions to cover the huge bad debt caused by this incident. Aave’s risk control partner, LlamaRisk, has proposed two main disposal plans:

1. Loss sharing across the entire network: distribute the losses uniformly across all deployed chains on Aave, with ETH lenders on the Ethereum mainnet bearing 1.54% of the loss;

2. Isolated loss management: limit losses to the layer-2 network where the hacker used rsETH for collateralized borrowing, with ETH lenders on Mantle potentially facing up to 71% loss.

All these estimates were made before Arbitrum froze 30,766 ETH from the hacker; the actual final loss could be significantly lower.

Additionally, Aave does not rule out using treasury funds to cover some bad debts, and Mantle’s official team has confirmed they are formulating asset recovery and compensation plans.

Personally, I hope the final solution can maximize user protection, achieving zero or minimal losses. For a long time, Aave has been a benchmark DeFi application in low-risk yield farming, but this security incident has seriously damaged its reputation.

After the incident, many voices have criticized the industry, claiming that top protocols are repeatedly experiencing failures and that DeFi is heading toward decline.

I do not agree with this view. Looking back at its development history, DeFi has faced multiple major crises but has always managed to repair, iterate, and restart to recover.

DeFi will not disappear because of this, but the entire industry must face the issues: before pursuing innovation and profits, security must become the top priority.