Short answer: in February 2025 the exchange Bybit lost about $1.5 billion in ether from a multisignature cold wallet. According to the Safe Ecosystem Foundation, the attack was made possible by a compromised Safe{Wallet} developer machine, which led to “the proposal of a disguised malicious transaction”. Several authorised signers approved it. The lesson for everyone: the interface can lie, so the decisive check must happen on the signing device.

What happened

  • The target: a Bybit cold wallet secured by a Safe (formerly Gnosis Safe) multisignature contract, which required several people to approve each transaction.
  • The entry point: according to Safe’s statement of 28 February 2025, a compromised developer machine. Reporting by BleepingComputer and The Block, citing the forensic investigations, describes malicious JavaScript served by Safe{Wallet}’s infrastructure that targeted Bybit’s wallet specifically.
  • The trick: the signers saw what looked like a routine transfer in the web interface. The transaction actually handed control of the wallet to the attacker.
  • The loss: about 401,000 ETH, roughly $1.4 to $1.5 billion depending on the price used, one of the largest thefts in the history of crypto.
  • What was not broken: Safe said external reviews found no vulnerabilities in its smart contracts or in the source code of its frontend and services. Bybit said its own infrastructure was not compromised.

Why multisig did not save them

Multisignature protects against one compromised key. It does not protect against several honest people approving the same falsified request. If every signer trusts the same web page, the attacker only needs to compromise that page. When the signing devices cannot display the real effect of a contract call in readable form, the signers are effectively signing blind.

Lessons for individuals

  1. The browser is not a trusted display. Only your hardware wallet’s screen is.
  2. Do not approve contract calls you cannot read on the device. If the device shows raw data, stop and verify through an independent tool or source.
  3. Diversify verification. In a multisig, signers should check independently, ideally with different software, not by trusting one shared screen.
  4. Be wary of “routine” requests. Attackers disguise malicious transactions as the most boring operation possible.

Frequently asked questions

Was Safe’s smart contract hacked?

According to Safe’s statement, no: external reviews found no vulnerabilities in the Safe smart contracts or in the source code of its frontend and services. The attack went through a compromised developer machine and a disguised transaction proposal.

Did the signers use hardware wallets?

Public reporting describes signers approving what the interface presented as a routine transaction. The general lesson holds whatever devices were used: if a device cannot show the real effect of a contract call, the signer is trusting the interface.

What can an individual learn from an exchange hack?

That the weakest link is often the screen you trust. Verify on your hardware wallet, be suspicious of urgent or “routine” requests, and never sign data you cannot read.

Sources

Facts as reported by the parties and the press in 2025; investigations may have refined some details since. Educational content, not financial advice.