Some exploits have a villain, a protocol and a post-mortem. This one, so far, has none of the three. On Sunday, October 4, 2026, a vault on Base that no project has claimed lost about $6 million in wstETH in under an hour — after someone with control of its 3-of-7 multisig quietly changed who was allowed to borrow from it.
Nobody knows who the seven signers are. Nobody knows yet whether the drain was theft, a rescue or a misconfigured permission. What is clear is how fast a single admin change can turn into a multi-million-dollar outflow when there is no delay between the decision and the money moving.
Key facts
- On October 4, 2026, an unclaimed vault on Base lost roughly 1,783 wstETH, worth more than $6 million, after a whitelist change added a new borrower.
- Two whitelist edits at 08:52–08:53 UTC added a newly deployed contract as an approved borrower, according to Crypto Briefing.
- Blockaid flagged the exploit at 09:21 UTC at about $2.02 million; roughly 40 minutes later losses had passed $6 million.
- The vault is owned by a 3-of-7 Safe multisig created about 324 days earlier; its signers are unidentified and no protocol has claimed it.
- About $31.7 million remained in the vault after the drain, per Crypto Briefing.
- Base and Aave V3 were not compromised; Aave honored valid receipt tokens.
What happened to the Base vault on October 4?
On October 4, 2026, two whitelist edits added a freshly deployed contract as a borrower on an unclaimed Base vault. That contract then borrowed 1,783.067 aBaswstETH — Aave receipt tokens — and redeemed them through Aave for about 1,783 wstETH, worth over $6 million, in six outflows over roughly 25 minutes.
The first outflow was a small test transfer, according to Crypto Briefing. Security firms PeckShield, CertiK and Exvul each arrived at the same total. Bitcoin.com News reported that Blockaid raised the alarm at 09:21 UTC, when losses stood at about $2.02 million; around 40 minutes later they had passed $6 million.
| Time (UTC, Oct 4) | Event |
|---|---|
| 08:52–08:53 | Two whitelist edits add a newly deployed contract as a borrower |
| Shortly after | First outflow, a test transfer; six outflows over ~25 minutes in total |
| 09:21 | Blockaid flags the exploit at ~$2.02M |
| ~40 minutes later | Losses pass $6M (~1,783 wstETH) |
Some of the funds reportedly moved through Lido’s Base-to-Ethereum bridge, which has a seven-day settlement period, according to Crypto Briefing.
Who controls the vault, and was it really a hack?
The vault is controlled by a 3-of-7 Safe multisig created about 324 days before the incident, and none of its seven signers have been identified. No protocol has claimed the vault. Whether this was a hack is not confirmed: compromised keys, a white-hat rescue or faulty permissions are all still possible explanations.
A few details make the picture murkier. The multisig had executed no transactions in the 25 days before the drain, Crypto Briefing reported. Then, within a minute, it pushed two whitelist changes — and within half an hour the vault was $6 million lighter. There also appears to have been no timelock, the standard safeguard that forces admin changes to wait before they take effect.
About $31.7 million was still in the vault afterwards. If this was a rescue, that remaining balance is the obvious open question. If it was theft, it’s a reminder that whoever can sign can also drain.
Three signatures and no delay: that’s the whole security model.
Were Base and Aave hacked?
No. Neither Base nor Aave V3 was compromised, according to Bitcoin.com News and Crypto Briefing. The drain happened at the level of the vault’s own permissions: a borrower was whitelisted, and Aave did exactly what it is designed to do by honoring valid receipt tokens presented for redemption.
That distinction matters. The underlying lending market and the layer-2 behaved correctly. The weak point was a layer built on top of them — a contract whose behavior could be changed by a small group of anonymous keyholders with no waiting period.
Why timelocks matter
A timelock doesn’t stop a bad admin change, but it gives everyone time to see it coming. Monitoring firms like Blockaid flagged this exploit about half an hour after the whitelist edits. With even a short delay on those edits, that alert would have landed before the first outflow rather than in the middle of the drain.
What does this mean for self-custody and swap users?
Every vault, pool and yield wrapper is someone else’s set of keys. When you deposit, you trust not just the code but whoever can change it. In this case, that was seven people nobody can name.
The simplest defense is to keep funds you aren’t actively using in a wallet only you control — ideally in cold storage — and move them only when you need to. When you do need to swap, a non-custodial route avoids parking funds anywhere in between. SwapZilla doesn’t hold your coins: you send to a provider’s deposit address and receive at your own address, without accounts or token approvals — see how to swap without connecting a wallet. If you’re moving funds across chains, our guide to swapping safely across bridges covers the rest.
Final thoughts
The October 4 drain is still missing its most important facts: who signed, why, and where the remaining $31.7 million is headed. Until those are known, it can’t honestly be called a hack, a rescue or an accident.
What it can be called is a clean example of admin-key risk. A dormant 3-of-7 multisig, two whitelist edits, no timelock, and $6 million gone in about 25 minutes. Audits of the underlying protocols didn’t help, because the underlying protocols weren’t the problem. The keys were.