# Iron Fish Ends Network Support After Merkle-Tree Inflation Bug

Published 2026-11-01 · Crypto.world Team · Crypto.world

> Iron Fish traced abnormal wIRON activity to a Merkle-tree flaw that could let a holder spend the same shielded note more than once. Its committed fix changes consensus roots and is intended for a new chain, not an in-place repair of existing histories.

[Iron Fish](https://github.com/iron-fish/ironfish/blob/master/docs/merkle-note-position-alias-postmortem.md) says it will end all remaining support for its network after tracing abnormal wIRON activity to an inflation bug in the blockchain itself. But the team’s postmortem describes a fix that cannot be applied to the existing chain without a separate consensus-upgrade design: the corrected Merkle-tree rules produce different roots from the historical rules.

The team announced the support withdrawal on October 30, 2026, after reviewing ChainPort bridge logs with the operator. Its technical postmortem, marked as a draft, explains how the flaw let a holder spend the same shielded note more than once. Iron Fish has not disclosed an amount created through the exploit or a verified value moved through the bridge.

## A copied Merkle branch broke the nullifier check

Iron Fish’s postmortem traces the design choice to an early Merkle-tree prototype from February 2019. When a right-hand branch was absent, the old implementation filled it by copying the populated left branch. That made the two inputs to the parent hash identical, so switching their order could preserve the tree root while changing the hidden position used in the spend proof.

A note’s position also feeds into its nullifier, the public identifier intended to stop that note from being spent twice. With the copied branch, a holder with the note’s spending authority could prove the same output at an alternate, unpopulated position and produce a different nullifier. The network’s duplicate-nullifier check then saw a new identifier rather than a repeat spend.

> “The fix replaces copied branches with a fixed empty value throughout tree root and witness construction.”
>
> The authors of Iron Fish’s technical postmortem write in the [postmortem](https://github.com/iron-fish/ironfish/blob/master/docs/merkle-note-position-alias-postmortem.md)

The postmortem distinguishes this mechanism from taking another user’s note: an attacker must control the spending authority and know the note contents. The flaw can allow repeated spending of a note the attacker owns, creating outputs whose combined value exceeds the original note. Iron Fish says its investigation reproduced the code-level alias, but its technical report does not identify which historical transactions used it.

## The patch does not repair the existing chain

Changing how empty branches are handled changes consensus roots, including roots associated with past chain states. The patch is intended for a new chain with a newly generated genesis and does not make existing mainnet or testnet histories compatible with the new rules, according to Iron Fish’s postmortem. It contains no activation-height rule, legacy-proof migration, or historical accounting plan.

That boundary helps explain why a code fix and a network recovery are separate problems. Preserving an existing chain would require a distinct consensus-upgrade design and security review, the document says. The completed local tests do not amount to an independent cryptographic audit or make old chains valid under the changed roots, the document also says.

## Audits covered the code, but historical attribution remains open

The affected code path fell within the scope of internal and external audits, Iron Fish said, yet reviewers did not catch the flaw. The postmortem describes additional adversarial tests of tree positions and spend proofs after the fix, while cautioning that those tests establish behavior of the corrected implementation—not which private witness was used in past transactions.

That caveat matters because Iron Fish’s privacy design keeps note positions hidden. Public proofs, roots and distinct nullifiers alone cannot establish that a particular historical transaction exploited the alias, the postmortem says. The team’s announcement links the abnormal activity involving wIRON and ChainPort to the blockchain bug, but the document leaves transaction-by-transaction attribution and the total amount unresolved.

The patch had already been committed and the postmortem included in the pull request, Iron Fish’s announcement said. Ending the team’s support does not establish that independently operated nodes have all stopped; the announcement concerns the withdrawal of remaining project support. The concrete technical limit is in the postmortem: the committed root-rule fix is for a new chain, not an in-place repair of Iron Fish’s existing history.

## Sources

1. Original source [Iron Fish on GitHub](https://github.com/iron-fish/ironfish/blob/master/docs/merkle-note-position-alias-postmortem.md) github.com

Other pages read

2. [Iron Fish on X](https://x.com/ironfishcrypto/status/2116278010802586090) x.com

## Related stories

- [Paper Finds a Tor Path to Monero Transaction Origins](https://crypto.world/news/paper-tor-path-monero-transaction-origins)
- [Researchers Find 9 of 10 AI Models Raised Their Own Circle Wallet Caps](https://crypto.world/news/ai-models-raised-circle-wallet-spending-caps)
- [b14g Reports Unauthorized Withdrawal of 97M CORE](https://crypto.world/news/b14g-unauthorized-withdrawal-97m-core)
- [Binance Sets Phased Withdrawal Restart After $6.3B Hack](https://crypto.world/news/binance-phased-withdrawal-restart-6-3b-hack)
- [Cardano Assigns CIP-0206 to Quantum-Resistant Plutus Signatures](https://crypto.world/news/cardano-cip-0206-quantum-resistant-plutus-signatures)

## Explore

[More articles](https://crypto.world/news)

---

Canonical: https://crypto.world/news/iron-fish-ends-network-support-merkle-tree-inflation-bug
