Security

Did a French Bank Really Lose $133 Million in Crypto?

An investigator traced over $133 million in crypto withdrawals to French bank Crédit Mutuel Arkéa. The bank confirmed an attack. Here is what is known.

Traced on-chain before the bank said a word: a tile crosses one of four windows on a teal route toward a lit green exit, the public on-chain trail. Tess reads a receipt at lower right.

Something unusual happened over the first days of November 2026: an independent on-chain investigator traced a pattern of withdrawals for two days before naming who he believed they belonged to, and a separate analytics firm had already confirmed a large piece of the trail on its own. The investigator is ZachXBT, a researcher with a long record of tracing stolen crypto. The name he landed on is Crédit Mutuel Arkéa, a French banking group, and the Crédit Mutuel Arkéa crypto theft he describes runs upward of $133 million. The bank has since confirmed an attack, but not the amount. Here are the questions worth asking before taking that number, or the order in which it came out, at face value.

What did ZachXBT actually find?

On Monday, November 2, 2026, ZachXBT reported a theft from centralized exchange accounts that he first put above $106 million. He then found another $27 million in withdrawals, pushing his estimate past $133 million, and said the figure could rise once over-the-counter trades are counted. On Wednesday, November 4, he named Crédit Mutuel Arkéa as the suspected victim on his Telegram channel. That is an investigator's estimate, built from watching wallet activity, not a figure any regulator or the bank has confirmed. He also said he had a high degree of confidence that North Korea's Lazarus Group was not behind it, citing differences in how the funds moved. CoinDesk's November 4 report on the allegation walks through the timeline as it stood that day.

Does anyone else back up the number?

Partially, and this is the detail that makes the story harder to wave away. Bitquery, a separate blockchain analytics firm, traced $108.8 million in USDC withdrawn from accounts at Kraken, Crypto.com and OKX over 21 hours on November 1 and 2, then followed it through hundreds of wallets, with much of it converted into ether, which no central issuer can freeze. Bitquery did not verify the full $133 million or identify the account holder. Two different trackers converging on the same multi-exchange withdrawal pattern, in the same window, by different methods, is a stronger signal than one investigator's claim alone, even with the top-line total still unconfirmed.

What did the exchanges do about it?

Crypto.com told CoinDesk it had acted to block funds linked to the incident and that all user funds remained secure. OKX said it had spotted unusual activity on an institutional customer account, froze it and contacted the client, with no sign of wider impact. Kraken did not immediately respond to CoinDesk. Two exchanges freezing specific flagged funds is a concrete response, and it is also not a denial that something happened.

What has the bank actually said?

When CoinDesk asked on November 4, Crédit Mutuel Arkéa confirmed the attack, saying it had enabled fraudulent transactions and was now under control. A spokesperson said the incident has no financial consequences for clients and that their interests are fully protected. The bank said it had temporarily suspended some transactions, notified the relevant authorities and was still assessing the consequences. It did not say how much was taken, or whether the money was first held in euros or in crypto. Some context makes the story sharper: in May, the French financial paper L'Agefi reported that the bank had applied for a license under the EU's MiCA rules to offer crypto services.

So who surfaced this story first, and why does that matter?

An outside investigator did, using public blockchain data and his own judgment about what looked wrong, two days before the bank confirmed anything, and only then because a reporter asked. That part of the story has nothing to do with the final dollar figure and everything to do with how custody gets checked. The accounts in question were not visible to the public. The withdrawals were not something a customer could watch happen. The reason anyone outside the bank and the exchanges knows about this at all is that someone read public, on-chain records and found a pattern worth naming.

What does a signed, published record change about that?

Not the amount of money at risk in any given system, and not whether a bank or an exchange gets targeted. What it changes is who has to go looking, and how. NexusPool's Glass Ledger signs an hourly receipt for the work it counts from every connected rig, covering the shares, the difficulty they were served at, and the window they landed in, against a key NexusPool publishes for anyone to check, instead of waiting on a company statement that may or may not arrive. That record exists whether or not anything ever goes wrong, which is a different posture from confirmation arriving after an outside investigator forces the question. NexusPool's design also removes one kind of account to go looking at in the first place: the block pays your address directly, with Payout Preflight showing the exact coinbase transaction before any block is found, so there is no pooled balance sitting in an exchange-style account for a credential issue or an access compromise to drain. More on how that payout path works is on NexusPool's page explaining its mining technology.

None of this is a claim about how anyone got access to the accounts involved, which remains unconfirmed, and it is not an accusation against the bank, Kraken, Crypto.com or OKX, none of which has been shown to have done anything wrong. It is not investment advice, and it does not claim NexusPool's core software is published for public review, which it is not. A published record also changes nothing about a miner's odds of finding a block, which are set by hashrate against network difficulty alone, the same at every pool, and holding or mining bitcoin carries risk that no record removes.

The biggest question this story raises is not how much money moved. It is why the public learned that anything had moved from an independent investigator, two days before the institution involved said a word.

Trust nothing. Verify who has to tell you something went wrong, and who found out on their own.