On April 18, 2026, a single attacker executed the largest DeFi exploit of the year and the damage did not stop at one protocol.

So, what happened

In the early hours of April 18, 2026, an attacker drained 116,500 rsETH worth approximately $292 million from KelpDAO's cross chain bridge. The stolen tokens were then deposited as collateral on Aave V3's mainnet instance and used to borrow a massive volume of Wrapped Ether. Because the rsETH collateral was by that point completely unbacked, the resulting debt positions cannot be liquidated through any normal mechanism. This is the reason why several investors on Aave where unable to liquidate their positions.

This left Aave's WETH reserve holding over $236 million in irrecoverable bad debt. The AAVE token itself dropped approximately 10% within hours of the news breaking. Both Aave V3 and Aave V4 rsETH markets were immediately frozen. SparkLend, Fluid, Compound V3, and Euler were also impacted.

This is now the largest DeFi exploit of 2026, surpassing even the $285 million Drift Protocol breach that occurred just weeks earlier on April 1.

What is rsETH and why does it matter?

To understand the full scope of this incident, you should first understand what rsETH actually is and why it was accepted as collateral on a protocol as significant as Aave in the first place.

KelpDAO is a liquid restaking protocol built on top of EigenLayer. The concept is straightforward. Users deposit ETH into KelpDAO, which routes it through EigenLayer to earn additional yield on top of standard Ethereum staking rewards. In return, the user receives rsETH. This is a tradeable receipt token representing their restaked position.

Liquid restaking tokens like rsETH became enormously popular over 2024 and 2025 because they allow users to earn layered yields while keeping their capital liquid and usable across DeFi. rsETH was deployed across more than 20 blockchain networks. This includes Arbitrum, Base, Linea, Blast, Mantle, and Scroll, with cross chain movement handled by LayerZero's Omnichain Fungible Token (OFT) standard.

The bridge at the centre of this exploit held the rsETH reserves that backed wrapped versions of the token on every Layer 2 deployment. That is what made it such a high value target.

How the exploit was executed

The attack was methodical and clearly pre planned. Here is the sequence of events as reconstructed from on chain data.

Firstly, hours before the exploit, the attacker funded the wallets involved using Tornado Cash's 1-ETH mixing pool. This is a technique commonly used to obscure the origin of funds and delay on chain attribution. This preparation was then followed, at precisely 17:35 UTC on April 18, by the attacker calling the `lzReceive` function on LayerZero's EndpointV2 contract. This function is the entry point through which cross chain messages are delivered and processed on the destination chain. The attacker crafted a spoofed message that tricked LayerZero's cross chain verification layer into believing a legitimate transfer instruction had arrived from another network. That spoofed message triggered KelpDAO's bridge contract to release 116,500 rsETH directly into an attacker controlled wallet.

In plain terms, the attacker forged what looked like a valid cross chain order, and the bridge obeyed it.

Now, what was left was to deal with the stolen loot. This was done by depositing the stolen collateral on Aave V3. This is where the exploit became a systemic event rather than an isolated theft. Instead of simply pocketing the rsETH and attempting to swap it for ETH, which would have been difficult given the token's size relative to available liquidity; the attacker deposited the stolen rsETH directly into Aave V3 on mainnet as collateral. Against that collateral, the attacker borrowed a substantial volume of WETH.

From Aave's perspective, it was processing what appeared to be a standard collateralised borrow. Aave's contracts were not compromised, and no bug in Aave's code was exploited. The protocol simply did what it was designed to do, that is to accept a recognised token as collateral and release a loan against it. The moment the bridge was drained, rsETH ceased to be backed by any real underlying ETH. The collateral underpinning those Aave borrow positions became worthless. Because the health factors on those positions effectively collapsed simultaneously and because the rsETH collateral cannot be redeemed for anything of value, Aave's standard liquidation mechanism cannot function. There is no asset to seize that would cover the debt.

KelpDAO's emergency multisig froze the protocol's core contracts 46 minutes after the initial drain, at 18:21 UTC. Two further attempts to drain an additional 40,000 rsETH, worth roughly another $100 million. This was blocked after that emergency pause, each reverting at 18:26 and 18:28 UTC.

The damage across the DeFi ecosystem

The contagion spread rapidly, illustrating in real time how interconnected DeFi protocols actually are.

Aave froze rsETH markets on both V3 and V4 deployments. Aave confirmed that its own smart contracts were not exploited and that the origin of the problem was entirely external. However, the bad debt already accumulated in its WETH reserve is real and must be addressed. Aave's Umbrella system which is a safety backstop that replaced the legacy Safety Module in late 2025 is designed precisely for this scenario. Users who have staked aWETH in the Umbrella vault face the possibility of automatic slashing to cover the deficit. Once that slashing cycle completes, WETH suppliers may regain partial withdrawal access, though full recovery is not guaranteed and some depositors may face a haircut.

What others are doing

SparkLend and Fluid both froze their rsETH markets as a precaution. SparkLend reported zero rsETH exposure, crediting its more conservative risk framework.

Lido Finance paused deposits into its earnETH product, which carries rsETH exposure. The core Lido staking protocol and the stETH token were confirmed to be completely uninvolved.

Ethena, the stablecoin issuer, temporarily shut down its own LayerZero bridges from Ethereum mainnet as a precautionary measure, despite having no rsETH exposure and maintaining over 101% collateralisation at the time.

Upshift paused deposits and withdrawals to its High Growth ETH and Kelp Gain vaults pending investigation.

On chain investigator ZachXBT flagged the active exploit publicly at 19:44 UTC, alerting the broader community. Solidity developer 0xQuit was among the first to issue a direct warning to WETH suppliers on Aave V3 to withdraw immediately. Aave founder Marc Zeller echoed that urgency.

Why Aave's contracts were not the problem

This distinction matters and deserves clear explanation. Aave V3 did not have a bug. Nobody exploited a vulnerability in Aave's code. What happened is that Aave accepted rsETH as collateral in good faith, under the assumption that rsETH would remain fully backed by real underlying assets. That assumption was shattered by the KelpDAO bridge exploit.

This is the fundamental risk of DeFi composability. Protocols are built to interoperate, which is what makes them powerful and capital efficient. A token minted by one protocol can be deposited in another, used as collateral in a third, and bridged across chains via a fourth. And this can all in a single transaction. But that same interconnection means that a breach anywhere in the chain can cascade instantly into every connected protocol. There is no circuit breaker, no waiting period, and no committee vote.

The attacker did not find a bug in Aave. The attacker found a bug in KelpDAO's bridge, and then used Aave's composability against it.

What This Means for Liquid Restaking Tokens

This exploit raises uncomfortable but necessary questions about how liquid restaking tokens are evaluated as collateral assets on DeFi lending protocols. rsETH was whitelisted on Aave because it represented a large and growing share of Ethereum's locked value and offered legitimate yield opportunities. The risk models used to assess it presumably included smart contract risk on rsETH itself. What they may not have fully modelled is the cross chain bridge risk. That is the possibility that the reserve backing rsETH on multiple Layer 2 networks could be drained through a messaging layer exploit.

This is not a flaw unique to rsETH. Any token that is bridged across chains and used as collateral carries a compounded risk profile. That is a smart contract risk on the underlying protocol, bridge risk on the cross chain messaging layer, and then all of the standard risks associated with lending protocol participation.

Saturday's incident marks KelpDAO's second major disruption within a year. In April 2025, a bug in its fee contract caused excess rsETH minting that led to a temporary pause, though no user funds were lost in that instance. The outcome this time is markedly different.

Final thoughts and conclusion

If you currently supply WETH on Aave V3, monitor the protocol's official communications closely. The Umbrella slashing mechanism, if triggered, will affect aWETH stakers in the Umbrella vault. Withdrawal availability may be limited until the deficit is resolved.

If you hold rsETH on any Layer 2 network, your tokens are now backed by a bridge reserve that has been partially drained. The ability to redeem for ETH at par depends on whether Kelp DAO can recover stolen funds or unwind enough of its EigenLayer positions to honor withdrawals

If you use LayerZero-powered bridges for any asset, watch for updates on the root cause analysis. Ethena's precautionary bridge pause and the broader community reaction suggest that trust in LayerZero's message verification security is under scrutiny until a full post-mortem is published.

The DeFi ecosystem is more resilient than it was in 2022. But resilience is not the same as immunity. One bridge exploit, properly weaponised, just created a quarter billion dollar hole in the world's largest DeFi lending protocol. That is still a serious problem.

Disclaimer: This is a developing story. All figures are based on on-chain data and reporting available at the time of writing. This article is for educational and informational purposes only and does not constitute financial advice.