Fraud Proof

A fraud proof is the enforcement tool that makes the "optimistic" part of an optimistic rollup safe to trust. Because the rollup posts new state roots to Ethereum without proving them correct upfront, the network needs a way to catch and punish a sequencer that lies. A fraud proof is that mechanism: a piece of evidence, verified onchain, showing exactly where an operator's claimed result diverges from the correct one.

In practice, most modern systems use an interactive, or "bisection," fraud proof rather than one giant recomputation. A challenger and the defending operator repeatedly split a disputed transaction batch in half, narrowing disagreement down step by step until only a single computational instruction is in question. That one step is then replayed directly on Ethereum, which is cheap enough to verify but still inherits Ethereum's security. Arbitrum's BoLD protocol, live on mainnet since February 2025, formalized this into a permissionless "all-vs-all" tournament with a fixed time bound, replacing an earlier system that relied on a whitelisted set of validators.

If a fraud proof succeeds, the fraudulent state root is discarded, the correct one is restored, and the dishonest party's posted bond is slashed, with a share often going to the challenger as a reward. This is why users typically wait out a challenge window, commonly around seven days, before withdrawing funds to Ethereum: the delay gives honest watchers time to submit a fraud proof if needed. Fraud proofs contrast with the validity proofs used by zk-rollups, which mathematically guarantee correctness at the moment of submission rather than relying on a post-hoc challenge.