On April 22, 2026, Arbitrum’s Security Council did something that most people in crypto assumed was impossible on a major Layer-2 network: it unilaterally froze 30,766 ETH — worth approximately $71 million at the time — belonging to the wallet linked to the KelpDAO bridge exploit. The funds were frozen without a community vote, without a governance proposal, and without any advance notice to the broader ecosystem. The action protected users. It also forced a conversation that the Ethereum community has been quietly avoiding for years: what does decentralization actually mean on Ethereum’s Layer-2 networks?
What Happened — and the KelpDAO Connection
The chain of events began on April 18, 2026, when an attacker exploited a flaw in KelpDAO’s rsETH bridge. The exploit involved minting fake rsETH tokens and using them as collateral to borrow approximately $292 million in real assets from Aave’s Ethereum markets. When news of the hack spread, a bank run followed — over $6 billion left Aave within 24 hours, collapsing the protocol’s total value locked from $45.8 billion to roughly $29.6 billion in a single day. For a detailed breakdown of how that exploit unfolded, see our coverage of the $290M KelpDAO exploit and its implications for Ethereum’s DeFi security.
A portion of the stolen funds — 30,766 ETH — had moved onto Arbitrum. The attacker’s wallet was identified on-chain, and the Arbitrum Security Council made the decision to freeze it. The freeze was executed using the Security Council’s emergency powers, which allow a 9-of-12 multisig of council members to take unilateral action in situations deemed to constitute a security emergency. No governance vote was required. No timelock applied. The funds were frozen within hours of the decision being made.
Aave founder Stani Kulechov publicly acknowledged the action, posting on X that the Security Council’s recovery of the $71 million could meaningfully reduce the unpaid debt left by the exploit. “Every decision we are making is aimed at an orderly return to normal market conditions and the best possible outcome for everyone involved,” Kulechov wrote. From a user protection standpoint, the freeze was unambiguously positive. From a decentralization standpoint, it raised questions that are harder to answer.
How the Arbitrum Security Council Actually Works
The Arbitrum Security Council is a 12-member multisig that holds emergency powers over the Arbitrum One and Arbitrum Nova networks. Its members are elected by ARB token holders through a governance process, and they serve six-month terms. The council has two modes of action: a 9-of-12 threshold for emergency actions (which can be executed immediately, without a timelock), and a 7-of-12 threshold for non-emergency actions (which are subject to a governance delay).
The emergency powers exist precisely for situations like this one. The Arbitrum governance documentation explicitly contemplates scenarios where rapid action is needed to protect users from active exploits. The KelpDAO situation — where identified stolen funds were sitting in a known wallet on Arbitrum — fits that definition. The Security Council used the powers it was designed to use, in the way it was designed to use them.
| Action Type | Threshold Required | Timelock | Community Vote | Example Use Case |
|---|---|---|---|---|
| Emergency action | 9 of 12 members | None | Not required | Freeze exploit funds, patch critical bug |
| Non-emergency action | 7 of 12 members | Yes (governance delay) | Not required | Protocol upgrades, parameter changes |
| Standard governance | ARB token vote | Yes | Required | Major protocol changes, treasury decisions |
What makes this case different from a routine security patch is the nature of the action. Freezing funds in a specific wallet is not the same as patching a smart contract vulnerability. It is a targeted intervention against a specific address — one that, from the protocol’s perspective, is indistinguishable from any other address. The Security Council did not freeze a buggy contract. It froze a wallet because of what that wallet’s owner had done on a different network.
The Decentralization Debate — Both Sides of the Argument
The crypto community’s reaction split almost immediately along predictable lines. On one side: the freeze was the right call. Stolen funds are stolen funds. The attacker had no legitimate claim to the ETH. The Security Council acted within its documented powers to protect users who had been harmed. If a 12-person multisig with elected members can prevent $71 million in stolen assets from disappearing forever, that is a feature, not a bug.
On the other side: the ability to freeze any wallet, for any reason, by a 9-of-12 multisig is a form of censorship that contradicts the foundational promise of permissionless finance. The argument is not that the attacker deserves to keep the funds. The argument is that a system where 9 people can freeze any address is not meaningfully different from a bank — and if it is not meaningfully different from a bank, the entire value proposition of building on a “decentralized” Layer-2 needs to be re-examined.
“Every decision we are making is aimed at an orderly return to normal market conditions and the best possible outcome for everyone involved.”
Stani Kulechov, founder of Aave, April 22, 2026
The deeper tension is about what “decentralized” means in practice versus in theory. Arbitrum is not Ethereum mainnet. It is a Layer-2 that settles to Ethereum, but its execution environment is controlled by Offchain Labs and governed by the Arbitrum DAO and Security Council. The security guarantees are different. The trust assumptions are different. Most users who interact with Arbitrum do so without fully understanding those differences — and the KelpDAO freeze has made those differences impossible to ignore.
What This Means for Layer-2 Governance Across Ethereum
Arbitrum is not alone in having these powers. Most major Ethereum Layer-2 networks — including Optimism, Base, and zkSync — have some form of privileged admin key or security council that can take emergency actions without a full governance vote. The specific thresholds and timelock structures vary, but the underlying architecture is similar: there is a trusted set of actors who can intervene in emergencies. This is a deliberate design choice, made because fully decentralized governance is too slow to respond to active exploits.
The question the Arbitrum freeze raises is whether “emergency powers” should extend to freezing specific user wallets, or whether they should be limited to actions that affect the protocol itself — like pausing contracts, patching bugs, or rolling back state. Those are meaningfully different categories of intervention, and the crypto community has not yet reached consensus on where the line should be drawn. The Arbitrum governance documentation does not draw that line explicitly, which is part of why this case has generated so much debate.
For Ethereum’s long-term trajectory, the governance question is inseparable from the scaling question. Layer-2 networks are where most Ethereum users actually live — where transactions are cheap, where DeFi is accessible, where the ecosystem is growing fastest. If those networks carry trust assumptions that are closer to a regulated financial institution than to a permissionless blockchain, that changes the risk profile for everyone building and using them. For more on how Ethereum’s broader roadmap addresses these tradeoffs, see our coverage of Vitalik’s 4-year plan for Ethereum, including the push toward zkEVM and stronger decentralization guarantees.
The Path Forward for Arbitrum
Arbitrum’s governance team has acknowledged that the freeze raises questions that need to be addressed through the DAO. The most likely outcome is a governance proposal that clarifies the scope of the Security Council’s emergency powers — specifically, whether wallet freezes fall within those powers, and if so, under what conditions. Some community members have proposed requiring a public on-chain record of the reasoning behind any emergency action, which would at least create accountability even if it does not change the underlying power structure.
The frozen ETH itself remains in limbo. The Security Council has the technical ability to freeze the funds, but the legal and governance process for what happens next — whether the funds are returned to affected users, transferred to a recovery fund, or handled through some other mechanism — is still being worked out. The Arbitrum DAO will almost certainly need to vote on the disposition of the funds, which means the story is far from over.
What is clear is that the KelpDAO freeze has accelerated a conversation that was already building in the Ethereum ecosystem about the governance maturity of Layer-2 networks. The Ethereum community has spent years debating the decentralization of the base layer. The Layer-2 layer has largely escaped that scrutiny — until now. For context on how Ethereum’s DeFi security landscape has been evolving, our analysis of the KelpDAO exploit and its broader implications provides the full picture of what triggered this chain of events.
Key Takeaways
The Arbitrum Security Council’s decision to freeze $71 million in stolen ETH was legally defensible, technically within its documented powers, and arguably the right call for the users who were harmed. None of that makes the underlying governance question go away. A 9-of-12 multisig that can freeze any wallet on the network is a form of centralized control — however well-intentioned, however democratically elected, however carefully documented.
The real question is not whether the Security Council made the right call in this specific case. It almost certainly did. The real question is whether the architecture that made that call possible is compatible with the long-term vision of a permissionless, censorship-resistant financial system. That is a question the Ethereum ecosystem — not just Arbitrum — needs to answer, because the same architecture exists in various forms across every major Layer-2 network.
The most honest answer, right now, is that Ethereum’s Layer-2 networks are not fully decentralized — and most of the people building and using them already knew that. What the Arbitrum freeze has done is make that fact visible to a much wider audience, at a moment when the stakes are high enough that the conversation can no longer be deferred.
