A $292 million exploit of Kelp DAO's rsETH bridge on April 18, 2026 — attributed to North Korea's Lazarus Group — has produced DeFi's first industry-wide coordinated bailout. The coalition, operating under the name "DeFi United," has raised more than $300 million in pledged ETH from Aave, Consens...
"The issue was beyond Aave. It was about restoring the whole state of DeFi, avoiding contagion and ensuring that the whole ecosystem overcomes this incident." — Stani Kulechov, Founder, Aave
A $292 million exploit of Kelp DAO's rsETH bridge on April 18, 2026 — attributed to North Korea's Lazarus Group — has produced DeFi's first industry-wide coordinated bailout. The coalition, operating under the name "DeFi United," has raised more than $300 million in pledged ETH from Aave, Consensys, Lido, EtherFi, Mantle Network, LayerZero, and others to restore rsETH's economic backing and cover approximately $200 million in bad debt on Aave.
Simultaneously, Arbitrum's 12-member Security Council executed an emergency freeze of 30,766 ETH (~$71 million) linked to the attacker — the largest asset freeze ever performed by a Layer 2 governance body. A DAO temperature-check vote on releasing those funds to the DeFi United recovery Safe opened May 1, drawing 16.9 million ARB in favor within the first hour and zero opposing votes.
The episode is structurally significant. It represents the first time a cross-protocol coalition has assembled a nine-figure fund to absorb exploit losses, the first time a Layer 2 security council has frozen and proposed to redirect stolen funds through formal governance, and the first time DAO token holders are being asked to adjudicate the disposition of assets seized from a state-sponsored attacker. The precedent — whether it holds or fractures — will shape how DeFi handles systemic crises for years.
At 17:35 UTC on April 18, 2026, an attacker exploited Kelp DAO's LayerZero V2-powered cross-chain bridge to release 116,500 unbacked rsETH on Ethereum — approximately 18% of rsETH's 630,000-token circulating supply. The attack was not a smart contract vulnerability. According to Chainalysis's forensic report, the attacker compromised Kelp's internal RPC nodes and launched a DDoS against external nodes, forcing a failover that fed false data to LayerZero's verification layer.
The core failure: rsETH was configured with a single Decentralized Verifier Network (DVN) — the LayerZero Labs DVN — in a 1-of-1 setup. No second verifier was required to confirm cross-chain messages. The attacker forged an inbound LayerZero packet that appeared to confirm a token burn on Unichain. No such burn occurred. Kelp's Ethereum bridge contract released the tokens based on the single fraudulent verification.
Within minutes, the attacker deposited 89,567 rsETH as collateral across Aave (Ethereum Core and Arbitrum), Compound, and Euler, borrowing approximately $190 million in WETH, wstETH, and other assets against the unbacked positions.
Kelp's incident response team detected the anomaly and paused contracts on Ethereum and its Layer 2 deployments, blacklisted the attacker's addresses, and engaged emergency security collective SEAL-911. Those actions blocked a second attempt to drain an additional 40,000 rsETH (~$95 million) using another forged packet. Without the pause, total losses could have exceeded $387 million.
LayerZero attributed the attack to the TraderTraitor sub-group of North Korea's Lazarus Group. Combined with the Drift Protocol exploit on April 1, Lazarus Group drained more than $575 million from DeFi protocols in 18 days through two structurally different attack vectors.
The exploit's impact extended far beyond Kelp DAO. Because rsETH was used as collateral across multiple lending protocols, the sudden creation of 116,500 unbacked tokens threatened the solvency of any protocol holding rsETH-backed positions.
Aave's exposure was the most severe. The attacker's 89,567 rsETH deposit generated roughly $190 million in borrows. Aave's governance immediately froze rsETH markets, set loan-to-value ratios to zero, and halted new borrowing against the asset. According to Aave's April 20 incident report, estimated bad debt ranged from $124 million (if losses were socialized across all rsETH holders, producing a ~15% depeg) to $230 million (if losses were isolated to Layer 2 rsETH holders).
Users responded by withdrawing capital. Aave saw approximately $6 billion in net asset outflows in the days following the exploit, according to reporting by CoinDesk. SparkLend and Fluid also froze rsETH markets. The rsETH token, which trades across more than 20 networks, faced backing uncertainty on every chain where it existed.
By April 26, on-chain analytics firm Arkham reported that Aave had raised approximately $160 million of the roughly $200 million needed to cover the bad debt — approximately 80% of the gap.
At 11:26 PM ET on April 21, Arbitrum's Security Council — a 12-member elected body with emergency powers — executed a 9-of-12 supermajority vote to freeze 30,766 ETH (approximately $71 million) linked to the attacker on Arbitrum One. The funds were moved to a governance-controlled intermediary wallet.
The freeze recovered roughly a quarter of the total assets drained. According to the Security Council's forum announcement, the action was taken under existing emergency provisions in Arbitrum's governance constitution.
Steven Goldfeder, co-founder of Offchain Labs (Arbitrum's development company), described the deliberation process: "The default was do nothing." The decision to act was not automatic — it required the council to weigh the precedent of intervention against the risk of inaction.
Under Arbitrum's governance architecture, the freeze is not final. The Security Council cannot unilaterally redirect the funds. Any movement of the frozen ETH requires a full DAO governance vote, meaning the disposition of $71 million rests with ARB token holders.
This is the largest asset freeze ever executed by a Layer 2 governance body and the first time frozen exploit proceeds have been routed to a formal DAO vote for disposition.
Within days of the exploit, Aave organized "DeFi United" — a coalition of DeFi protocols and ecosystem participants — to pool capital and restore rsETH's economic backing. By April 27, pledges exceeded $300 million.
Key contributors, according to CoinDesk and Bitcoin.com reporting:
| Contributor | Pledge | |---|---| | Aave DAO | Up to 250,000 ETH (allocation from treasury) | | Stani Kulechov (personal) | 5,000 ETH | | Consensys / Joseph Lubin | Up to 30,000 ETH | | Mantle Network | 30,000 ETH backstop | | LayerZero | $23 million | | Lido, EtherFi, and others | Additional commitments (amounts vary) |
The recovery mechanism works through a 2-of-3 Gnosis Safe controlled by signers from Aave Labs, Kelp DAO, and on-chain security firm Certora. The plan calls for converting committed ETH into rsETH in tranches and transferring the tokens to the affected lockbox contract to restore the bridge's backing ratio.
If the Arbitrum DAO vote passes, the 30,766 frozen ETH would also flow into this Safe, adding $71 million to the recovery pool.
The structure is notable for its formality. DeFi United is not an ad-hoc telegram group — it has named signers, a defined Safe architecture, a technical restoration plan published on Aave's governance forum, and a return mechanism if the recovery plan fails. It is, in effect, an organized creditor committee operating on-chain.
On May 1, 2026, the Arbitrum DAO opened a Snapshot temperature-check vote on the proposal to release the 30,766 frozen ETH to the DeFi United Gnosis Safe. Within the first hour, approximately 16.9 million ARB tokens were cast in favor. Zero votes were cast against.
The temperature check runs until May 7. If it passes, the proposal advances to an on-chain binding vote on Tally.
The vote raises a structural question that DeFi governance has not previously confronted at this scale: whether ARB token holders have both the technical authority and the normative legitimacy to direct the movement of funds that originated from an exploit of a separate protocol (Kelp DAO), were laundered through a third protocol (Aave), and were frozen on a fourth platform (Arbitrum).
No existing DeFi governance framework cleanly addresses cross-protocol asset recovery. The resolution Arbitrum's community reaches will likely be cited as precedent in future incidents.
The Security Council's freeze has divided the DeFi community along predictable but sharp lines.
The intervention case: Allowing a state-sponsored attacker to retain $71 million in proceeds — when the technical capability to prevent it exists — amounts to subsidizing North Korean weapons programs in the name of ideological purity. The Security Council acted within its constitutional mandate, and the DAO vote ensures the broader community ratifies the decision.
The anti-intervention case: Connor Howe, CEO of cross-chain infrastructure project Enso, stated: "Crypto protocols are not that different from centralized platforms or banks if a small group of people can freeze funds." If 12 people can freeze any wallet on Arbitrum, the network's censorship-resistance guarantees are conditional, not absolute.
The debate extends beyond Arbitrum. Justin Sun publicly contrasted Tron's architecture against Arbitrum's, claiming greater decentralization — though Tron's own governance concentration has been widely documented. The broader question is whether Layer 2 networks, which inherit Ethereum's security but maintain independent governance structures, can credibly claim to be permissionless when they retain emergency override capabilities.
Neither side has a clean answer. The economic reality is that $71 million in frozen North Korean proceeds is a policy outcome that most jurisdictions would mandate by law. The philosophical reality is that the power to freeze those funds is the same power that could, under different political conditions, freeze funds belonging to dissidents, competitors, or politically disfavored actors.
The exploit surfaced a systemic issue in LayerZero's cross-chain messaging architecture. Kelp DAO's bridge was configured with a 1/1 DVN setup — a single verifier with no redundancy. According to Kelp DAO, this was LayerZero's default configuration. According to LayerZero, Kelp was advised to adopt multi-DVN architecture but chose not to.
Kelp DAO's position, per CoinDesk reporting: LayerZero's quickstart guide and default GitHub configuration pointed to a 1/1 DVN setup. Approximately 40% of protocols on LayerZero were using the same configuration at the time of the exploit. Kelp maintained a direct communications channel with LayerZero since July 2024, during which no specific recommendation to change the rsETH DVN configuration was produced.
LayerZero's position: Best practices around DVN diversification were communicated to Kelp DAO. The responsibility for configuration rested with the application team.
In response, LayerZero announced it would stop signing messages for any applications using single-validator setups and is forcing a "security migration" requiring all OApps (Omnichain Applications) to adopt multi-DVN architectures.
The configuration dispute is relevant beyond the blame game. If 40% of LayerZero-connected protocols were running 1/1 DVN setups at the time of the Kelp exploit, the attack surface was not an edge case — it was a systemic default. LayerZero's forced migration to multi-DVN is an implicit acknowledgment that the prior configuration standard was insufficient for the value being secured.
The Kelp DAO exploit and its aftermath mark a structural inflection point in how DeFi handles systemic failures. For the first time, the industry assembled a nine-figure coordinated response — with named participants, a defined custody structure, and a technical restoration plan — rather than leaving affected users to absorb losses individually.
The mechanism worked, or is working: $160 million of $200 million in needed coverage was raised within eight days. The Arbitrum freeze recovered a quarter of stolen assets. The DAO vote is proceeding with apparent consensus.
But the episode also exposed uncomfortable realities. A $292 million exploit succeeded because a default configuration left a bridge secured by a single verifier. A Layer 2 security council demonstrated that it can freeze any wallet on its network with a 9-of-12 vote. And the recovery mechanism — however effective — is functionally a bailout, with the costs distributed across ecosystem participants rather than borne by the protocol whose configuration failed.
DeFi United may become the template for future crisis response. It may also become the precedent that regulators cite when arguing that DeFi already has identifiable responsible parties and bailout mechanisms — and should therefore be regulated as such.
The data does not suggest which interpretation will prevail. Both are supported by the facts.