Aztec Connect $2.19M ZK-Rollup Theft: Settlement Boundary Bypass Creates L1/L2 State Discrepancy
On June 14, 2026, a deprecated Aztec Connect RollupProcessor contract was exploited via a ZK-rollup settlement boundary bypass. The attacker crafted a gap between numRealTxs and decoded_slots, allowing 31/32 ZK public-input slots to be committed without L1 settlement verification.
Aztec Connect’s RollupProcessorV3 relies on assumptions that each public-input slot is either verified at the L1 layer or constrained by the ZK circuit (publicValue == 0). The reported flaw breaks this logic by making the L1 settlement loop process fewer slots than the ZK proof commitment range, creating an opening for “unsupported” minting on L2.
Attack details (single atomic transaction):
- Attacker EOA: 0x0f18d8b44a740272f0be4d08338d2b165b7edd17
- Exploit tx: 0x074ec931…aee1
- Mint phase: 7 processRollup calls (Rollup #13277–13283) produced inflated L2 balances.
- Withdrawal phase: 7 processRollup calls (Rollup #13284–13290) cashed out the inflated L2 balances back to L1 assets.
- Reported net theft: ~ $2.19 million from the RollupProcessor’s L1 pool.
Reported cash-out assets include DAI (270,513.054), wstETH (167.890), yvDAI (4,873.857), yvWETH (16.570), LUSD (9,273.734), yvLUSD (359.047), and ETH (908.987).
Tracing: as of 2026-06-15, all stolen funds were reportedly transferred to the attacker EOA and remained intact, with the intermediate contract holding no remaining funds.
For traders, this is another reminder that ZK-rollup security depends on strict L1/L2 boundary consistency. Aztec Connect-related risk may raise scrutiny on rollup settlement logic and legacy contract exposure.
Neutral
This report flags a highly specific ZK-rollup settlement-boundary logic flaw that enabled an ~ $2.19M theft, but it is not evidence of a broad, systemic failure across the entire market. The most direct market relevance is risk pricing around rollup settlement verification and legacy contract exposure.
In the short term, incidents like this often trigger a sentiment hit for the affected ecosystem (traders may reduce exposure to related rollup/layer-2 narratives until mitigations are confirmed). Similar past patterns—where state-derivation or boundary-check bugs led to “proof verified but settlement not validated” outcomes—tend to cause temporary volatility in L2-related tokens and in on-chain activity metrics.
In the long term, the clearer takeaway for traders is that security reviews focusing on L1/L2 boundary consistency, calldata decoding trust boundaries, and independent verification of ZK public inputs can become a selection filter. That can support a more durable rebound for well-audited systems, while leaving the reputational overhang for projects with deprecated or legacy holdings.
Overall, because the event appears localized to a specific contract and does not automatically imply wider rollup protocol-level consensus risk, the expected market impact is best categorized as neutral rather than broadly bullish or bearish.