Bitget says private keys were not compromised, so how did the attacker actually move the funds out? How is this different from a typical exchange hack?
The classic pattern in most exchange hacks is that an attacker obtains the private keys to a cold or Hot Wallet directly, then simply signs a transfer transaction and moves assets out, which is why exchange security typically centers on making sure private keys never leak. Bitget case was different: the attacker never touched private keys at all, instead compromising a third-party security product the exchange used to manage its wallet infrastructure, gaining internal network access, then using that access to forge transaction data so the spoofed transfer instructions looked like legitimate transactions that had passed through the normal internal approval process.
That means what the attacker bypassed was not the layer of who holds the keys, it was the verification mechanism that confirms whether a transfer was actually authorized in the first place. That is also why this incident is worth paying attention to: exchange security spending tends to concentrate on protecting private keys, multisig, Cold Storage, hardware security modules, but if a third-party system connected to wallet infrastructure has its own weakness, an attacker can get the system to misjudge a forged transaction as legitimate without the private keys ever moving at all.
The CEO said the company suspects North Korea. Can that statement be taken as established fact? Does it carry the same evidentiary weight as past North Korea attribution cases backed by detailed technical reports?
Not quite. The evidence Bitget CEO Gracy Chen cited on the livestream included the pattern of asset conversion, with some assets rapidly swapped into ETH, the IP addresses used by the attacker, and some funds flowing into wallets previously linked to Lazarus Group, these are all common categories of clues used in the crypto industry when assessing attacker identity, and they are indeed the same categories of evidence that have supported successful North Korea attributions in past cases. But based on what public reporting currently shows, this attribution rests primarily on Bitget internal team assessment and the CEO own public statement during a livestream, with no formal report from an independent forensic security firm, an on-chain analysis company, or a government agency surfacing to corroborate it.
That does not mean the North Korea attribution is wrong, the evidence types cited do carry real reference value, and North Korea linked hacking groups documented activity in the crypto space over recent years is a well-supported historical pattern. But until an independent verification report emerges, the more careful stance is to treat this attribution as the exchange current best guess rather than an already-confirmed conclusion.
If the protection fund has to absorb a 76% hit, does that mean Bitget is close to being unable to cope? If another attack happened again, would the fund still be enough?
Looking at the fund being wiped 76% in one event on its own is understandably alarming, but it is worth distinguishing between the fund being depleted and the exchange itself becoming insolvent, Chen noted that beyond the protection fund, Bitget holds more than $1 billion in proprietary assets, and the exchange most recent pre-breach proof-of-reserves report showed an aggregate Reserve Ratio of 135% across 19 assets, meaning the exchange books already held more in assets than total user balances before this happened. Even with three-quarters of the protection fund wiped out, as long as the company own assets and overall reserve position do not show a major gap, it should in theory still be able to make good on this compensation commitment.
The more important question, though, is what happens next time: a buffer fund designed to handle occasional unexpected incidents just lost three-quarters of its value in a single event, meaning the room it has left to absorb another incident of similar scale has shrunk substantially. Public information has not yet clarified whether Bitget will replenish the fund immediately following this incident, to what size, or whether any new minimum reserve threshold will be set, these remain open questions worth watching but without an answer yet.
If I have assets on Bitget, what should I do right now? When will withdrawals resume?
Withdrawals are currently paused for a security review, and public reporting has not surfaced a clear timeline for when they will resume, Bitget has committed to publishing a full incident report within 24 hours of the breach, the more practical approach is to keep watching the exchange own official announcements and the CEO follow-up statements rather than relying on secondhand summaries circulating on social media. Deposits and spot trading have continued operating normally since the incident, meaning the exchange has not gone into a full shutdown, a contrast with the handful of past cases where an exchange simply closed services or collapsed without warning after being hacked.
Worth sitting with is that this incident itself reinforces a habit worth having before any withdrawal pause happens at all: not keeping more assets than you can afford to lose sitting long-term in a single exchange Hot Wallet, particularly when an exchange security perimeter may not just cover its own systems but also the third-party security vendors it depends on, since the breach in this case came through an outside security product Bitget used to manage its wallet infrastructure, not a system Bitget built itself.
At 18:31 UTC on September 24, Bitget exchange security systems flagged an abnormal internal fund movement, which was quickly confirmed as a breach of the exchange hot and warm wallets, with roughly $351.6 million in assets moved out, the largest exchange hack of 2026 so far and one that ranks among the ten largest crypto thefts on record. Bitget CEO Gracy Chen responded quickly with a livestream on X, stating that user funds are safe, your account balances are accurate, and your assets are protected, and announcing that the loss would be fully covered by Bitget User Protection Fund, which holds more than $464 million. Cold Storage wallets were entirely unaffected, deposits and spot trading continued operating normally, and only withdrawals were temporarily paused pending a security review.
The method behind this attack differs from what most people picture when they imagine an exchange getting hacked. Chen stated explicitly on the livestream that Private Key compromise has been ruled out, and she described the attack as comparable to forged withdrawal slips appearing inside a bank: the attacker did not steal any private keys at all, instead compromising a third-party security product Bitget used to manage its wallet infrastructure, using that foothold to obtain internal network credentials, then using that access to spoof transaction data so that forged transfer instructions appeared to pass through the normal internal approval process as legitimate transactions, triggering no alerts. In other words, what the attacker broke was not the defense protecting where the money sits, it was the verification layer that confirms whether a given transfer was actually authorized in the first place. On the asset side, the attacker converted roughly $100 million into ETH, on top of about $85 million already held in ETH, while $157.5 million in XRP and $7 million in TRX remained unconverted, with USDT and USDC holdings also targeted for conversion.
Chen said on the livestream that the company suspects North Korea, citing the pattern of asset conversion, with some assets rapidly swapped into ETH, the IP addresses used by the attacker, and transfers into wallets previously linked to North Korea Lazarus Group. It is worth noting, though, that this attribution currently rests primarily on Bitget own internal assessment and the CEO own public statement during a livestream, public reporting has not surfaced a formal attribution confirmation from an independent forensic security firm or a government agency, which is not quite the same evidentiary standing as past North Korea attribution cases backed by full technical reports from on-chain analysis firms, such as some earlier attacks on other exchanges. If this attack is ultimately confirmed as North Korea linked, it would push North Korea cumulative 2026 crypto theft total past the billion-dollar mark.
Bitget User Protection Fund currently holds assets valued at more than $464 million, denominated primarily in 5,500 bitcoin. If this $351.6 million loss is absorbed entirely by that fund, it would consume roughly 76% of the fund total value in a single event, a ratio worth sitting with on its own: a buffer fund meant to handle unexpected incidents just had three-quarters of it wiped out by one incident, meaning the cushion available for whatever comes next has shrunk dramatically. Chen response to this was that beyond the protection fund, Bitget holds more than $1 billion in proprietary assets, and that customer funds remain backed on a 1:1 basis, with the exchange most recent pre-breach proof-of-reserves report showing an aggregate Reserve Ratio of 135% across 19 assets, though that report predates the breach and does not reflect the exchange current standing, and public information has not clarified whether any minimum reserve floor exists if the fund is drawn down heavily.
Bitget has committed to full compensation, and the exchange itself has not halted operations or frozen accounts, a meaningful difference from past cases where an exchange simply collapsed after a hack and users lost everything, such as some earlier exchange failures. But a compensation commitment is still a commitment, not a completed action, withdrawals were paused immediately after the incident and no official timeline for resuming them has been given. For users holding assets on Bitget, the more practical stance is to keep watching official follow-up announcements while understanding that a protection fund existing and a protection fund being large enough to cover this loss plus whatever comes next are two different things, particularly once a single event has already consumed three-quarters of it.