The KelpDAO exploit has become one of the most consequential DeFi security incidents of the year, not only because of the immediate loss, but because of how quickly the damage spread across lending markets and liquidity venues. According to the source report, an attacker used a forged cross-chain message on April 18 to drain 116,500 restaked ETH, valued at roughly $292 million. Within 48 hours, the shock reportedly contributed to more than $13 billion in total value locked outflows across the broader DeFi ecosystem.
Cardano founder and Ethereum co-founder Charles Hoskinson used the event to argue that the industry’s dominant threat model is outdated. In his view, smart contract bugs are no longer the only, or even primary, danger. Instead, bridge verification failures and message authentication weaknesses are becoming a systemic risk, especially when stolen assets can be rapidly routed into money markets before emergency controls take effect.
A bridge exploit that became a lending contagion event
The source article says the attacker submitted a spoofed Layerzero message to the endpoint v2 contract linked to Kelp’s restake adapter. That forged packet falsely claimed Uni-Chain endpoint ID 30320 as its origin, and the message was accepted because Kelp’s cross-chain configuration relied on a one-of-one decentralized verifier network. In practical terms, that meant a single verification path became a single point of failure.
Hoskinson emphasized that this setup illustrates why bridges can be especially fragile. If verification depends on only one trusted pathway, compromising that pathway can unlock escrowed assets without attacking the escrow logic itself. That distinction matters because it shifts the industry conversation from code bugs in isolated smart contracts to architectural flaws in cross-chain trust assumptions.
The most damaging part of the exploit, however, came after the initial drain. Rather than immediately dumping the stolen assets on decentralized exchanges and crashing the token’s price, the attacker allegedly deposited the restaked ETH into lending protocols such as Aave. By borrowing liquid wrapped ether against that collateral, the attacker could leave with assets that were no longer directly linked to the theft, while the tainted collateral remained embedded inside lending markets.
This is the mechanism Hoskinson described as novel and dangerous: the hack did not stop at the bridge. It moved into credit markets, where it created the conditions for bad debt, liquidity stress, and a rapid loss of confidence. That is why the incident was portrayed not just as a theft, but as a contagion event.
Scale of the fallout across DeFi
The report cited in the source material says Llamarisk confirmed 83,471 ETH equivalent spread across seven attacker wallets on Ethereum mainnet and Arbitrum. It also laid out two broad resolution scenarios. In the first, losses would be socialized across all restaked ETH holders through a 15.12% haircut, leading to roughly $123 million in bad debt absorbed by Ethereum core reserves. In the second, losses would be isolated at the layer-2 level, repricing affected tokens to just 26.46% backing and concentrating about $230 million in bad debt across networks such as Mantle, Arbitrum, and Base.
The wider liquidity impact was severe. Aave alone reportedly experienced between $6.6 billion and $8.45 billion in outflows. Wrapped ETH pools on Arbitrum, Base, Mantle, Linea, and Plasma approached full utilization, effectively making withdrawals difficult or impossible during the period of peak stress. At least nine DeFi protocols were described as directly affected, including Compound, Morpho, Lido, Ethena, Pendle, Euler, Beefy, and Lombard Finance.
These numbers underscore why the market reaction was much larger than the initial theft size. A hack worth around $290 million would already be significant, but a confidence shock that pulls more than $13 billion from DeFi within two days reveals how tightly connected bridge infrastructure, collateral systems, and liquidity markets have become.
Disagreement over responsibility
Three separate post-mortems were published by KelpDAO, Layerzero, and Llamarisk, and the source report notes that they do not fully agree on where responsibility lies. Layerzero announced on April 20 that it would no longer sign or attest messages for any application still running a one-of-one DVN configuration, and pushed for a migration to multi-verifier setups across the protocol.
Kelp, by contrast, maintained that Layerzero’s default cross-chain configuration had shipped with single-source verification across several major networks, including Ethereum, BNB Chain, Polygon, Arbitrum, and Optimism. The article further says Kelp claimed that roughly 40% to 50% of all Layerzero OFT applications may still be using the same one-of-one arrangement. If accurate, that would mean the security concern is not isolated to a single application, but potentially extends to a meaningful share of the broader ecosystem.
On-chain forensics reportedly suggested possible ties to the Lazarus Group, the state-linked hacking organization often associated with North Korea. Still, the source material is careful to note that no independent forensic firm had issued a formal attribution at the time, and the FBI had not publicly commented.
Why Hoskinson pointed to Cardano and Midnight
Hoskinson used the incident to contrast Cardano’s architecture with the restaking and bridge-heavy structures that amplified KelpDAO’s risk. He argued that Cardano’s liquid, non-custodial staking model reduces the need for the layered stack of staking, liquid staking, and restaking wrappers that can expand the attack surface in other ecosystems.
His more specific claim involved Midnight, Cardano’s privacy-focused sidechain. According to the article, Hoskinson said Midnight’s Nightstream protocol can package entire chain states into proofs that travel with cross-chain messages. In theory, that would allow recipients to verify whether a message reflects valid chain state before accepting it, potentially blocking forged messages at the verification layer rather than trying to contain damage after the fact.
He also highlighted multi-party computation, or MPC, as a way to make stronger verifier configurations easier to operate. Instead of relying on a one-of-one setup, protocols could move toward models such as two-of-three or five-of-seven verifier networks. The goal is straightforward: remove the single point of compromise that made the KelpDAO exploit possible.
Whether Midnight becomes a practical answer for the wider cross-chain industry remains an open question, but Hoskinson’s core argument is clear. If bridges remain necessary, then message verification must be far more robust, and risk controls must assume that stolen assets can be mobilized into lending markets almost immediately.
A warning for the next phase of DeFi security
Hoskinson also tied the incident to a broader concern about the role of artificial intelligence in offensive security. As referenced in the source article, he said advanced AI systems may allow attackers to scan codebases at a scale and speed no human review team can match, increasing the likelihood of discovering complex or emergent weaknesses across protocols and infrastructure.
That warning matters because the KelpDAO case was not simply a lesson about one protocol making one configuration mistake. It highlighted how modern DeFi risk now spans bridges, verifiers, restaking layers, lending markets, and emergency response systems. The article notes there was only a 46-minute window between the initial drain and Kelp’s emergency pause. In a fast-moving, highly composable ecosystem, that may not be enough time to prevent stolen collateral from spreading through credit markets.
For the DeFi sector, the lasting significance of the event may be less about the initial exploit than about the structure of the fallout. A forged message triggered a bridge failure; the bridge failure contaminated collateral; the contaminated collateral strained lending platforms; and that strain accelerated withdrawals across the ecosystem. In that sense, the KelpDAO incident became a live stress test for how decentralized finance handles cross-chain trust assumptions under pressure.
The central takeaway is that cross-chain security can no longer be treated as a niche infrastructure issue. As this incident shows, a weak verification model at one bridge can rapidly evolve into a market-wide confidence crisis. That is the vulnerability Hoskinson says Cardano and Midnight are designed to address, and it is also the challenge the rest of DeFi now has to confront more directly.