According to BlockBeats, on August 17th, Harmony Protocol announced a rollback plan for the ONE aberration event. The plan is to roll back Shard 0 to block 92,730,034 and Shard 1 to block 94,978,278, both corresponding to the time of August 11, 2026, at 23:25:37 UTC. Validators will use the replacement database for the corresponding blocks, and will subsequently regenerate blocks at heights of 92,730,035 and 94,978,279.
Harmony stated that the first confirmed instance of fraudulent issuance occurred in block 92,730,036 of Shard 0, therefore the previous block was chosen as a safety buffer. After evaluating solutions such as targeted burning, blacklisting, selective transaction replay, token migration, and direct database rollback, the team concluded that these solutions could potentially affect unrelated funds or cause inconsistencies in the chain state. Therefore, a fixed rollback window was ultimately chosen. Harmony disclosed that a fraudulent issuance wallet attempted 534 transfers of 5 billion ONE each within 106 seconds, with 477 successful transfers totaling 2.385 trillion ONE. The team stated that current fund flow tracking covers over 99.9% of fraudulent ONE paths, but "traceable to wallets or clusters" does not mean that specific individual identities can be confirmed, nor does it mean that all related funds can be safely destroyed.
Furthermore, Harmony has completed the full archiving of Shard 0 from blocks 92,730,035 to 92,871,662, encompassing 141,628 consecutive blocks, 109,126 regular transactions, and 315 staking transactions. The team stated that all blocks and regular transactions after the rollback checkpoint will be discarded, and they are currently working with trading platforms, cross-chain bridges, and law enforcement to assess the impact and develop solutions. Harmony claims that an independent third-party security firm has reviewed the incident and confirmed the forged issuance and the discovery of key fund flows. Validators must follow the official recovery process to perform the rollback.