About $6.3 billion left Binance on October 29, 2026. It is being reported as the largest crypto hack to date, and Binance says it has covered every customer loss.
Developers tend to read a story like that and ask "how did they get in?" I think the more useful question for anyone holding coins on a platform is a different one: "where exactly was my crypto sitting, and what had to go wrong for it to move?"
I build and run a crypto investing platform on my own, so I have had to answer that question for my own system. This post walks through the mechanics in plain English, then turns them into 7 checks you can run on any exchange before you deposit. No architecture tour. Just enough of the machinery to know what to look for.
The direct answer: on a custodial platform your balance is a database entry, the coins live in wallets the platform controls, and a hack is what happens when someone other than the platform gets those wallets to sign a transfer. What you get back afterwards depends on the platform's money, insurance, and terms, not on the blockchain.
Your balance is a row, not a wallet
When you deposit on an exchange, your coins are swept into wallets shared with every other customer. Your dashboard shows a number from the platform's database.
That means two separate things can be true at once: the database says you own 1 coin, and the wallet that should hold it is empty. The blockchain only knows about the wallet. The promise to you lives in the database and the terms.
This is why "not your keys, not your coins" gets repeated so often. It is not a slogan about ideology. It is a description of where the control sits.
Where the keys live decides how bad a hack can be
Platforms split funds across wallets with different trade-offs. You will see three terms.
Hot wallet. Keys are online so withdrawals can be automatic. Fast, and exposed. Most large thefts drain hot wallets.
Cold wallet. Keys are kept offline. Slow to move, much harder to reach remotely. The weak point becomes the moment a human brings them online to sign.
Multi-signature wallet. Moving funds needs several independent approvals, for example 2 of 3 or 3 of 3. One stolen key is not enough. The weak point becomes anything all the signers trust in common, such as the screen that shows them what they are approving.
That last point matters more than it sounds. In several recent incidents the cryptography held perfectly. The attackers did not break the keys. They changed what the signers thought they were signing. A multi-signature setup lowers the odds. It does not remove the humans.
What "the exchange covered it" actually means
After a theft, the blockchain shows coins gone from the platform's wallets. Every customer's database row still shows the old balance. The gap between the two is the platform's problem to fill.
It has four ways to do that, and history has examples of each:
- Fill it from company money. Binance in 2026 and Bybit in 2025 did this.
- Claim on insurance. Usually covers part, up to a limit.
- Shrink every row. Bitfinex in 2016 cut all balances by about 36%.
- Stop and go to court. Mt. Gox in 2014, with creditors waiting about a decade. Nothing on-chain decides which one you get. It is a business and legal outcome, which is why the checks below are mostly about documents and only partly about wallets.
What a hack looks like from the customer's side
The sequence is similar almost every time, and knowing it removes some of the panic.
First, withdrawals stop. This is usually the first thing customers notice, often before any announcement. The platform is not being evasive. It is closing the door while it works out which wallets are still under its control.
Second, your dashboard looks normal. Balances do not change, because the database was not what got robbed. This is the most misleading moment of the whole event. A normal-looking balance during a withdrawal freeze tells you nothing about whether the coins exist.
Third, the platform says which ending it is choosing. Sometimes within hours, sometimes after days. Until that statement arrives, nobody outside the company knows whether customers will be made whole.
Fourth, the freeze lifts in stages, network by network or asset by asset, as wallets are rebuilt and keys replaced. Even in the best case this takes days to weeks.
Two practical points follow. During a freeze, ignore anyone who contacts you offering faster access or recovery. Scammers work every major incident. And notice that the only real protection against the waiting is a decision you made earlier: how much you chose to keep on that one platform.
That is why everything useful happens before a hack, which is what the checks are for.
The 7 checks
1. Search the terms for who bears a loss. Look for "loss", "unauthorized" and "security incident". Note whether the platform accepts liability or disclaims it.
2. Get the insurance details in writing. Insurer, covered wallets and assets, limit. Three facts or it does not count.
3. Find the wallets on a block explorer. If the platform publishes addresses, open them. Check whether they are ordinary wallets or multi-signature contracts, and how many signatures are required. This takes two minutes and is the only check you can do without trusting anyone.
4. Ask whether customer assets are lent, staked, or reused. If yes, the coins may not be in those wallets at all on a given day.
5. Look for assets matched against liabilities. Wallet lists prove what the platform holds. Only an audit or attestation with a date shows whether that is enough to cover what customers are owed.
6. Read the withdrawal suspension clause and past incident write-ups. Every platform can pause withdrawals. Look for limits on that power, and for evidence of how the platform communicates when something is attempted against it.
7. Test the exit and lock your own account. Make one small withdrawal early. Use a unique password and an authenticator app. For an individual, account takeover is far more likely than a headline hack, and no platform's vault design protects a weak inbox.
What I built, and what it does not solve
Here is how my own platform, Wealtii, answers the same checks today.
The tokens behind every Wealtii fund sit in one public multi-signature vault on BNB Smart Chain. The address is published and anyone can inspect it. Moving funds requires all three signatures. There is no pool of customer assets being lent or staked, because I do not lend, stake, or rehypothecate user assets at all.
That design narrows the ways funds can move. It also has costs I should name. Three required signatures means three things must all work for a legitimate withdrawal, and custody still depends on the signers and on me. A multi-signature vault is a reduction in risk, not an absence of it.
And on the part that decides the ending after a hack, my answers are weaker than a large exchange's:
- No deposit insurance, investor-compensation scheme, or crypto insurance.
- No third-party audit or attestation yet.
- No published reconciliation of vault holdings against fund units outstanding yet.
- Not registered or licensed with any financial regulator, and no formal legal entity yet.
- One operator. I could not refill a large loss from company money the way Binance did. Last week I published the minute-by-minute log of a 105-minute automated attack on Wealtii that failed at every step. I am glad it failed. It is one data point, and it tested the login and support surfaces, not the vault.
If you want to run check 3 on my platform right now, the vault address and the rest of these statements are on the Wealtii transparency page. Do that before you believe anything else I have written here.
Digital assets are volatile and you can lose some or all of the money you invest. Multi-signature custody does not remove market, issuer, custody, liquidity, or regulatory risk. This is educational content, not financial, legal, or security advice.

Top comments (0)