Aave liquidates Kelp DAO attacker’s rsETH collateral as

Aave Labs has liquidated the Kelp DAO attacker’s remaining rsETH-backed positions on Ethereum and Arbitrum and transferred the proceeds to DeFi United’s Recovery Guardian multisig. The recovery effort now hinges on 30,765 ETH frozen on Arbitrum that is subject to a restraining notice as a DAO vote nears its close.

Key Takeaways

The attacker’s remaining rsETH-backed positions on Aave across Ethereum and Arbitrum have been liquidated, with collateral transferred into DeFi United’s Recovery Guardian multisig.
The April 18, 2026 exploit is pegged at $293 million and left Aave with more than $190 million in bad debt after stolen rsETH was used to borrow wETH.
Aave says the liquidation process did not impact user funds, and its Umbrella insurance mechanism was not used.
Arbitrum DAO has 30,765 ETH frozen in “legal limbo” under a restraining notice as a vote to release the ETH to DeFi United approaches its Friday close.

Aave Clears the Attacker’s Remaining rsETH Positions Across Ethereum and Arbitrum

Aave Labs has now forced the issue on the attacker’s remaining rsETH-backed borrow positions. The protocol liquidated those positions on both Ethereum and Arbitrum, then transferred the liquidated collateral to Recovery Guardian, a multisignature wallet managed by DeFi United.

Aave framed the liquidation as a “critical step” in the “DeFi United” recovery plan. Mechanically, this matters because it moves the recovery from an on-protocol risk problem into an execution problem. Once collateral is sitting in a controlled multisig, the market is no longer staring at an open question of when the attacker’s remaining positions might unwind and what that would do to liquidity.

Aave also drew a bright line around contagion optics. It said user funds were not affected by the liquidations, and that Umbrella, its automated bad-debt protection mechanism, was not used.

How the $293M Kelp DAO Exploit Turned Into $190M+ of Bad Debt on Aave

The exploit itself is dated April 18, 2026, and sized at $293 million. The second-order damage came from how the stolen asset was used.

The attacker used stolen rsETH as collateral on Aave and borrowed wrapped Ether (wETH) against it. When collateral quality is compromised but the borrow leg is a high-liquidity asset like wETH, the protocol can be left holding the bag if liquidations cannot fully cover the debt. That is the bad-debt pathway in one line.

In this case, the deficit was large. The incident left Aave with more than $190 million in bad debt, and it triggered a wave of withdrawals that hit the protocol’s total value locked (TVL). Aave’s TVL fell by nearly $12 billion in a week after the exploit.

What stands out here is the market structure lesson. Lending protocols can absorb isolated liquidations. They struggle when a compromised collateral type becomes a borrow engine for the most liquid base asset in the system. That is how a single exploit turns into a balance-sheet event.

Recovery Math: 13,000 ETH Unlocked, but DeFi United Still Needs the Final ~10%

Aave had telegraphed the immediate payoff from clearing the attacker’s remaining collateral. On April 28, it said that cleaning up the hacker’s collateral on Ethereum and Arbitrum would release 13,000 ETH worth nearly $30.2 million at then-current prices.

That liquidation milestone is now in the rearview, and the recovery math is tighter. Galaxy Digital research vice president Thaddeus Pinakiewicz put two separate “last mile” numbers on the situation.

First, he said DeFi United is only about 10% short of the ETH needed to restore the Kelp DAO restaked ETH (rsETH) token. Second, he said Aave is now “only 10% short of recovering from the bad debt” caused by the hack.

Those two statements are the same story told from different balance sheets. One is about restoring rsETH backing. The other is about Aave closing the hole left by undercollateralized borrowing. Either way, the liquidation reduces uncertainty around the attacker’s remaining on-protocol footprint, but it does not finish the job.

DeFi United is also waiting on additional commitments from Circle, Ethena, Frax, and Kraken-built Ethereum layer-2 Ink. Pinakiewicz said Aave needs these commitments to “get it over the line and plug the hole.” No amounts or timelines were disclosed, which keeps the final leg of the recovery probabilistic rather than scheduled.

Arbitrum’s 30,765 ETH Freeze Meets a Restraining Notice as the DAO Vote Nears Close

The biggest swing factor is not another liquidation. It is a frozen pile of ETH.

Pinakiewicz said 30,765 ETH is frozen by Arbitrum DAO and is in “legal limbo” after U.S. law firm Gerstein Harrow LLP filed a restraining notice to prevent Arbitrum DAO from redistributing the frozen ETH. Aave responded by filing an emergency motion to vacate the restraining notice.

In parallel, Arbitrum DAO members are voting on whether to release the frozen 30,765 ETH to the DeFi United fund. At the time of publication, more than 90% of voters were in favor, and the vote is set to close Friday.

The pattern worth noting is the sequencing risk. Even if the governance vote clears, the restraining notice can still dictate the practical timeline. Traders should treat the vote as a sentiment and coordination signal, not as a guarantee of immediate asset mobility.

On the market-health side, the post-exploit withdrawal shock appears to have moderated. DefiLlama data shows Aave’s TVL rose from a local low of $14.2 billion on April 26 to back above $15 billion, and net outflows eased over the past week. That does not erase the bad debt, but it suggests the acute bank-run phase has cooled.

Why This Legal-Governance Bottleneck Is the Real Risk Variable for Aave and rsETH

I view the liquidation as the easy part, even though it was operationally critical. It collapses one branch of uncertainty by removing the attacker’s remaining rsETH-backed positions from the live risk surface and pushing collateral into Recovery Guardian. That shifts the market’s focus from liquidation risk to execution risk.

Execution risk is now dominated by governance and legal process. The hard number is 30,765 ETH frozen on Arbitrum, explicitly described as being in “legal limbo” due to a restraining notice, with Aave filing an emergency motion to vacate. That is not a DeFi mechanism problem. It is a permissions problem.

Scenario one is the clean path. The Arbitrum DAO vote closes in favor, and the restraining notice is vacated on a timeline that allows redistribution. In that case, the recovery narrative tightens quickly because the largest discrete pool of stuck ETH becomes movable. Confirmation for this scenario is straightforward: a finalized vote outcome in favor, followed by a clear legal disposition that removes the transfer constraint.

Scenario two is the messy but survivable path. The vote passes, but the restraining notice remains in force longer than the market expects. That would keep the recovery in a holding pattern even as the attacker’s positions are already cleared. The tell here would be governance success without follow-through in asset movement, paired with continued reliance on the unspecified commitments from Circle, Ethena, Frax, and Ink.

Scenario three is the stress case. The vote fails or is effectively neutralized by the restraining notice, leaving the 30,765 ETH frozen with no near-term release path. In that world, the “only about 10% short” framing becomes less comforting because the last 10% is exactly where timelines and confidence can break. The invalidation point for the bullish interpretation of the liquidation milestone would be a prolonged inability to move the frozen ETH despite the vote process completing.

The TVL rebound from $14.2 billion on April 26 to above $15 billion, alongside easing net outflows, tells me the market has stopped treating this as an immediate solvency run. The core thesis is narrower now: the recovery’s outcome is increasingly gated by whether Arbitrum’s frozen 30,765 ETH can legally and operationally be released into the DeFi United recovery flow.

Sources

Cointelegraph