# ACryptoS shuts down after losing $665,000 across 77 vaults

> ACryptoS loses $665,000 across 77 vaults on five chains due to an input validation flaw, offers the attacker 10% and shuts down the project.
> By José Maldonado · DeFi · 2026-11-09

ACryptoS, a decentralized finance yield aggregator, announced in the early hours of November 9, 2026, that it is ceasing operations after an attacker drained about $665,000 from 77 vaults spread across five chains. The figure is modest for the size of the market and decisive for those who had money inside, because a contract that moves other people's funds was left without them for failing to check a basic piece of data before accepting a deposit.

A yield aggregator describes its work in one sentence and executes it for years: it takes deposits, spreads them across strategies that pay interest and returns the capital when someone claims it. Each vault is a contract with its own balance, and the shares are the proportional portion belonging to each depositor within that balance. ACryptoS operated with that logic across several networks at once, which turned a single flaw into a list of affected contracts.

The outcome is not the usual one either. The company did not speak of a pause to audit, but of closing the project, and the [ACryptoS notice from this morning](https://x.com/acryptosdao/status/2119479478749667555) asks all users to withdraw what remains while withdrawals are still open. Before shutting down the system, the company offered the attacker a 10% bounty to return the assets. On the morning of November 9, 2026, the clock is ticking for anyone with a position in those vaults who has not yet moved anything.

## Draining 77 vaults with minimal deposits

The attacker did not need a flash loan or to manipulate a reference price. They deposited minimal amounts of token into each vault and received in exchange the vast majority of that vault's shares. With control of the internal accounting, they withdrew the entire balance: the little capital they had put in and everything the previous users had contributed.

Seventy-seven vaults across five chains mean repetition, not improvisation. The same flaw was in every copy of the contract, so the technique worked the same in the first vault and the last.

![ACryptOS statement on X](https://content.observatorioblockchain.com/wp-content/uploads/2026/11/acryptos-x.webp)
*ACryptOS statement on X*

The mechanics of a share explain why a ridiculous amount is enough. If the contract calculates the participation from the incoming deposit and does not compare the result with the balance that already existed, a pittance of an input can return a disproportionate portion of the fund. A large number is not needed; what is needed is for the contract to accept whatever arrives without looking.

The fact that the damage is spread across five networks also changes how the incident reads. An attack on a single contract is contained with an emergency pause. Seventy-seven deployments force the operation to be repeated on each chain, and the result was $665,000 in a single night. The [double exploit of Tendril and CoreVault on BNB Chain](https://observatorioblockchain.com/en/defi/tendril-corevault-exploit-seguro-defi/) showed a similar pattern in November: small, fast attacks that no one stops in time.

## The lack of input validation in one sentence

The flaw fits in one line: the contract issued shares without checking that the amount received corresponded to the participation handed out. Validating an input means verifying that the data arriving is within reasonable bounds before using it to calculate a balance, and in these vaults that check did not exist on the deposit path.

These absences are not detected by reading quickly. A line that does not exist is not highlighted by any analysis tool, and human review tends to look at the code that is there, not the code that is missing.

The contrast with other thefts this month helps gauge the attacker's effort. Manipulating an oracle requires capital, a precisely constructed transaction and a window of seconds. Here a small deposit and a function that trusted the input were enough. The [arrival of RedStone prices on DogeOS](https://observatorioblockchain.com/en/defi/redstone-dogeos-defi-dogecoin/) describes the other side of the same problem, because a vault depends on external data and on its own as well.

There is a design lesson that reappears in every incident of this kind. Input validation is not a security layer added at the end, it is part of the calculation: without it, the contract accepts as truth any figure that reaches it, whether from a user, an oracle or another contract.

## The funds crossed chains and something remains on record

The stolen assets did not stay still. The trail passed through several interoperability protocols, bridges that move balances from one network to another, and each hop leaves two records: one on the source chain and one on the destination chain. The amounts and timings at both ends match, and that match is what makes it possible to reconstruct the path months later.

Tracing is not the same as recovering. A bridge documents the movement of money precisely and does not return a single dollar by itself. The chain says where the balance was, and says considerably less about who moved it.

The source addresses are public, and that gives a concrete advantage to whoever investigates. The initial deposit, the exact moment of the drain and the order in which the funds were moved can all be followed. The weak point of that trail is the conversion to money with custody: an exchange with identity verification forces someone to put a name to the receiving address.

The [security as a form of money and of time](https://observatorioblockchain.com/en/sociological-analysis/la-constelacion-de-la-seguridad-dinero-en-tiempos-de-hackeos/) is the framework in which these episodes are best understood. A protocol that holds other people's value does not only compete on yield, it also competes on the confidence that the balance will still be there tomorrow, and that confidence is lost faster than it is built.

## The 10% bounty and the withdrawal that remains open

ACryptoS tried to recover part of the money by offering the attacker a 10% bounty to return the funds. On $665,000, the offer is worth about $66,500, an amount that only makes sense if the thief prefers to collect without exposing the trail to someone with access to their movements identifying them. The company also asked users to withdraw the remaining funds while withdrawals remain open.

Bounties rarely work. They read as proof that the team is cooperating with the investigation and expose the fact that their best negotiating tool is money they no longer control.

The withdrawal notice has a practical consequence that the statement does not detail. If the available balance does not cover all positions, the order of arrival determines how much each user recovers: whoever withdraws first gets paid in full and whoever waits competes for what is left. In a closed protocol, haste stops being a character trait and becomes a financial decision.

The tally of affected vaults and of the five chains leaves a reasonable doubt about the total. The $665,000 figure is an estimate made with prices at the time of the attack, and in a portfolio spread across five networks the real value changes with every hour that passes.

## When should a project shut down instead of relaunching?

A relaunch makes sense when the flaw lives in a peripheral module, the team can pay for a full audit and the product retains users willing to return. Shutting down makes sense when the error is in the vault's core accounting, because then no patch can restore trust: every new deposit would be a bet on the same code that failed.

The size of the protocol does not decide the response, the nature of the flaw does. An aggregator with $665,000 in custody cannot fund the audit of five chains with the fees of a product that no longer inspires confidence.

There is also a market reason pushing toward closure. DeFi users tend to punish with exit, and a protocol that returns after a drain receives fewer deposits, pays more to attract them and competes with twenty clones that do not carry the history. ACryptoS chose not to take that path, and the decision is more honest than a relaunch with the same code.

For anyone with a position in these vaults, the checklist remains short: check whether withdrawals are open, how much is left in the contract and whether the team has published a destination address or a distribution plan. None of that brings back the money lost, and all of it prevents losing the next deposit in a protocol that has yet to explain its failure.
