Worked example · 2026-07
WEMIX — an attacker takes administrator privileges over the WEMIX$ stablecoin contract and mints 5.2M units out of nothing, forcing the network to shut its own bridges, DEX, and marketplace — WEMIX (Wemix3.0) — 2026-07-26
Summary
WEMIX is a blockchain network operated by the WEMIX Foundation, with WEMIX$ as its par-pegged stablecoin.
On 2026-07-26 at 18:17 KST (09:17 UTC), an unidentified attacker gained administrator privileges over the WEMIX$ stablecoin smart contract and minted approximately 5,225,525 WEMIX$ — an issuance worth about $5.22M at par — with no backing behind it.
The newly created tokens were routed through a decentralised exchange and swapped into 30,736 WEMIX and 724,198.27 USDC.e. The attacker then moved the USDC.e off WEMIX3.0 through bridges into Ethereum and BNB Smart Chain, where it was swapped into ETH and USDT and scattered across multiple wallets, some reaching centralised exchanges. Total assets taken reached approximately $6.25M; the $724,000 in bridged USDC is the portion confirmed as externally siphoned, which is why several outlets headline the smaller figure — the remainder consists of minted units and WEMIX proceeds that never left the network.
WEMIX responded with a comprehensive defensive lockdown: all bridges suspended — its published list covers every route connected to and from WEMIX3.0, including the Chainlink CCIP route and the PLAY Bridge — affected liquidity pools frozen, the WEMIX$ Module and PNIX DEX paused, and NFT marketplace trading disabled, while the team conducted emergency audits. The Foundation identified the attacker's wallets, coordinated with global exchanges to freeze assets, initiated a comprehensive contract audit, and tracked fund flows on-chain. WEMIX stated that its internal investigation into the root cause remained ongoing, with sequential updates promised on remediation and service-reactivation timelines.
This was the second major breach in 18 months. On 2025-02-28, attackers drained 8.65M WEMIX tokens worth $6.1–6.2M from the Play Bridge Vault using compromised authentication keys — Nile NFT-platform monitoring keys a developer had uploaded to a shared repository, obtained roughly two months before they were used. That is documented at examples/2025-02-wemix-play-bridge-vault-shared-repo-key-compromise.md and is a different vector reaching a comparable loss: an off-chain credential compromise, where the 2026 attack exploited contract ownership directly.
Timeline (UTC)
| When | Event | OAK ref |
|---|---|---|
| 2025-02-28 | Prior breach: 8.65M WEMIX (~$6.1–6.2M) drained from the Play Bridge Vault via Nile monitoring keys exposed in a shared repository; disclosed 2025-03-04, contributing to a June 2025 domestic delisting | (prior incident, different vector — see the 2025-02 example) |
| (undisclosed, prior) | Administrator privileges over the WEMIX$ stablecoin contract come under attacker control; method not publicly disclosed | T9.004 pre-positioning |
| 2026-07-26 09:17 | Attacker mints |
T9.004 / T5.003 exploitation |
| 2026-07-26 | Minted units routed through a DEX into WEMIX and USDC.e | T7.007 (conversion) |
| 2026-07-26 | Assets bridged to Ethereum and BNB Smart Chain, then deposited into centralised exchanges; ~$6.25M total, $724K bridged USDC confirmed externally siphoned | T7.003 / T7.002 |
| 2026-07-26 onward | WEMIX suspends all bridges, freezes affected liquidity pools, pauses the WEMIX$ Module and PNIX DEX, disables NFT marketplace trading; emergency audits begin | (operator containment) |
| after | Foundation identifies attacker wallets, coordinates with global exchanges to freeze assets, tracks flows on-chain | (operator response) |
What defenders observed
- Pre-event (mint authority as a single control). Contract ownership over a par-pegged asset is the highest-value privilege in the system, and it was reachable and immediately exercisable. Threshold control plus a timelock on the mint path is the mitigation; it converts a silent single-transaction issuance into a delayed, publicly-visible one that the operator can veto (M17, M22, M34).
- Pre-event (no issuance cap). 5,225,525 units in one action, against a peg, with no per-window bound. An issuance rate limit set well above legitimate operations and well below catastrophic ones costs nothing in normal running and caps the loss in this exact scenario (M04).
- At-event (total supply is the single best alert). A stablecoin's supply changing outside a scheduled issuance is a one-line monitoring rule with essentially no false positives. It would have fired at 09:17, before the DEX conversion, while the units were still on the home chain and inside the operator's own halt authority.
- At-event (the conversion path was the escape route). The attacker needed the DEX to turn unbacked WEMIX$ into WEMIX and USDC.e, then bridges to reach Ethereum and BNB Smart Chain. Every one of those hops is a venue the operator ultimately halted — but halted after the value had moved. The ordering is the lesson: containment authority over your own venues is only worth something if the trigger fires before the funds traverse them.
- Response (a proportionate but very expensive shutdown). Suspending bridges, pools, the DEX, and the NFT marketplace was the correct call, because an unbacked par-pegged unit converts at every venue honouring the peg. It also means a mint compromise costs far more than the minted amount — the network's entire economy stopped. That second-order cost belongs in any threat model that treats mint authority as "just another admin key."
- Response (second incident in 18 months, different vector). February 2025 was authentication keys; July 2026 was contract ownership. Fixing the first did not constrain the second, because they are different privileges reached by different means. Post-incident hardening scoped to the specific vector that was used leaves the neighbouring privileges untouched — the review question after a key compromise is which other authorities exist and how is each one reached.
Public references
[cryptotimeswemix2026]— The Crypto Times, "WEMIX Hacked Again: $6.25M Stablecoin Exploit Forces Network Shutdown" (2026-07-26 18:17 KST timing, ~$6.25M total and $724,000 bridged USDC, the ~5,225,525 WEMIX$ mint and administrator-privilege mechanism, the DEX → WEMIX / USDC.e → Ethereum and BNB Smart Chain → CEX path, the full lockdown of bridges / liquidity pools / WEMIX$ Module / PNIX DEX / NFT marketplace, the February 2025 8.65M WEMIX comparison, and the Foundation's exchange coordination): https://www.cryptotimes.io/2026/07/27/wemix-hacked-again-6-25m-stablecoin-exploit-forces-network-shutdown/[cryptoslatewemix2026]— CryptoSlate, "Compromised owner contract just let hackers print 5.2 million WEMIX stablecoins out of thin air, forcing a complete network freeze" (the compromised owner contract as the specific control, the 5.23M mint occurring outside the stablecoin's 100%-backing controls, and the network-freeze framing): https://cryptoslate.com/wemix-owner-breach-minted-5-23m-tokens-outside-its-100-backed-stablecoin-controls/[cryptonewswemix2026]— crypto.news, "WEMIX freezes bridges after owner-key breach mints 5.23M WEMIX$" (the swap into 30,736 WEMIX and 724,198.27 USDC.e, the movement of USDC.e off WEMIX3.0 into Ethereum and BNB Smart Chain with swaps into ETH and USDT scattered across multiple wallets, the suspension of every bridge to and from WEMIX3.0 including the Chainlink CCIP route and PLAY Bridge, and the ongoing root-cause investigation): https://crypto.news/wemix-freezes-bridges-after-owner-key-breach-mints-5-23m-wemix/[blockonomiwemix2026]— Blockonomi, "WEMIX Suffers Second Security Breach: $724K Stolen Through Unauthorized Token Minting" (independent confirmation of the second-breach framing and the $724K externally-siphoned figure): https://blockonomi.com/wemix-suffers-second-security-breach-724k-stolen-through-unauthorized-token-minting
Discussion
WEMIX sits alongside examples/2026-06-tesseradao-tsr-admin-key-unauthorized-mint.md and examples/2026-07-b2-network-staking-contract-upgrade-authority-seizure.md as the third administrative-privilege compromise OAK has documented in eight weeks, and the trio makes a coherent point about where 2026's losses are actually coming from. None of the three involved a broken cryptographic primitive, an exotic economic exploit, or a subtle logic bug. In each case a privileged function did precisely what it was written to do, for someone who should not have been able to call it. The industry's review capacity is heavily concentrated on the code path an ordinary user takes, and thinly spread across the code path an administrator takes — which is the path with no economic constraints on it.
What distinguishes WEMIX within that group is the peg. TesseraDAO and B² Network minted or extracted a governance-style token, and the market repriced it: the attacker's proceeds were bounded by available depth, and B²'s attacker in fact lost roughly 22% of the nominal value to slippage on exit. A par-pegged stablecoin removes that natural bound, because the peg is a standing commitment to honour each unit at face value at every venue that accepts it. That is why the response had to be a network-wide halt rather than a pool pause, and it is the reason a mint authority on a pegged asset should be governed more tightly than a mint authority on a floating one — the same compromise converts to realised value far more efficiently.
The recurrence is also worth recording carefully, and OAK should resist the shorthand that WEMIX "was hacked twice by the same failure." It was not. February 2025 was an off-chain compromise of authentication keys sitting in a shared repository, used to drain a bridge vault; July 2026 was an on-chain seizure of contract ownership, used to mint. They are neighbouring privileges reached by different routes, not the same one, and remediation of the first would not have touched the second. What the pair genuinely shares — argued at more length in examples/2025-02-wemix-play-bridge-vault-shared-repo-key-compromise.md — is that the shortest path in ran through a standing authority scoped against nothing, both times. Contributors documenting repeat incidents at a single operator should state explicitly whether the second event reused the first's vector, because "the same hole twice" and "a different hole in the same wall" imply very different things about the operator's security programme — and only the second is true here.