KETJU Research

← The Register

other

Arbitrum Bridge

Approved · limits
Max sleeve
15%
Reviewed
2026-08-17 · v1
Next review
2026-11-17
Research basis
Individual research
Protocol TVL, 30d
$3.34B +18%
Chains
Ethereum · sovereign, Arbitrum One · hybrid

The scheduled date is the outside bound. Kill criteria are checked every day, and a trigger reopens the memo that week.

APPROVED WITH LIMITS. Arbitrum’s canonical Nitro-based optimistic-rollup bridge has a clean exploit record on the bridge contracts themselves and a documented, non-theoretical emergency-response capability: on 2026-04-20, the 12-member Security Council froze about $71M in ETH linked to the Kelp DAO/LayerZero exploit, recovering roughly a quarter of the stolen value, then released it through a governance vote and federal court order once the legal process resolved. Non-emergency contract changes carry a real delay — roughly eleven days across an 8-day L2 and 3-day L1 window — a stronger structural check than Base’s zero-delay upgrade path, reviewed separately in this registry. The freeze episode cuts both ways: it protected users, but it also confirms 9 of 12 council members can unilaterally lock funds pending governance, a known and bounded concentration risk this cap prices rather than argues away.

The research file

Mechanism

Standard Nitro optimistic-rollup design. Deposits from Ethereum to Arbitrum are fast. Withdrawals are subject to a seven-day challenge period: the withdrawal is provisionally accepted while any party can submit a fraud proof, and funds are only claimable on Ethereum once the window closes without a successful challenge. This is structural to the rollup design, not a discretionary delay Arbitrum could shorten without changing its security model.

Control and governance

The Security Council is a 12-member multisig with two modes. Emergency Actions require 9-of-12 approval and can execute any action to protect chain integrity with no delay. Non-Emergency Actions also require 9-of-12 but must pass through the DAO’s AIP process, producing an 8-day L2 delay plus a 3-day L1 delay before deployment. Every emergency action requires a public transparency report afterward, and the DAO can remove Council members. This is a materially different structure from Base’s nested 2-of-2 operator-plus-council model with no delay on any upgrade path.

The April 2026 freeze: power exercised, not just documented

After the 2026-04-18 Kelp DAO/LayerZero exploit (reviewed separately in this registry’s `layerzero-v2` entry), the attacker used unbacked rsETH as Aave collateral to borrow about $230M in ETH. On 2026-04-20 the Security Council froze 30,766 ETH, about $71M, tied to the exploit, after consulting law enforcement, moving the funds beyond reach without a full governance vote. That recovered roughly a quarter of the total loss and is among the fastest such interventions on record. It also became entangled in a US federal court proceeding after parties tied to North Korea-related judgments sought a claim over the funds; Arbitrum delegates voted about 91% to approve release to an Aave-controlled recovery wallet, and a federal judge authorized the transfer on 2026-05-09. The native bridge contracts themselves were not exploited in this or any other identified incident; two July 2026 incidents sometimes associated with Arbitrum — the AFX Trade bridge hack and the Ostium oracle-key exploit — were confirmed to be third-party protocol failures unrelated to the canonical bridge.

Exit under stress

The seven-day challenge window applies uniformly with no officially expedited path; third-party liquidity bridges can front funds against a pending withdrawal faster, but carry a separate trust model and are out of scope for this canonical-bridge entry. A client relying on this bridge should plan around the full seven days for a guaranteed canonical exit, not assume same-day liquidity.

Comparison

Against Base’s canonical bridge (approved with limits in this registry): Base has zero delay on any upgrade once its 2-of-2 threshold is met, while Arbitrum’s non-emergency path carries a real, roughly eleven-day delay; both have a delay-free emergency path with a small elected group holding real unilateral power, and Arbitrum’s April 2026 freeze is direct evidence that power is exercised in practice, not theoretical. Neither is a reason to prefer one over the other outright — both carry the same class of governance-concentration risk, priced here at the same 15% cap.

Sources

The claims above trace to these. Where a number could not be independently verified, the thesis says so.

Inherited controls

The verdict above grades the protocol layer. Every position also inherits the asset it holds and the chain it settles on. The least safe layer sets the position’s grade, and the position table names which one that is.

ChainVerdictGradeControl constraint
EthereumApproved sovereign No sequencer, no upgrade key, no operator who can be compelled — rule changes require social consensus.
Arbitrum OneApproved · limits hybrid a single sequencer orders >99% of transactions and admin keys can upgrade bridge contracts on a ~7-day timelock.
The memo is public. The watching is the product: the terminal reads your clients’ wallets against this Register and flags the events above when they fire. $49 per advisor per month, first 14 days free. Start the trial.