The Bybit theft altered what signers saw, not the cryptography

Inherited archive entry · retained from the earlier edition. Its factual, legal and policy claims have not all been reverified for 19 September 2026. Research scope.

What the FBI and the wallet's foundation each recorded

On 21 February 2025, Bybit lost control of an Ethereum cold wallet during what was meant to be a routine transfer to a hot wallet. The FBI's public service announcement, issued five days later, states that approximately $1.5 billion in virtual assets was stolen and attributes the theft to North Korean actors it tracks under the name 'TraderTraitor'. A separate statement from the Safe Ecosystem Foundation, published 28 February 2025, narrows the point of entry: a forensic review found the attack reached Bybit's wallet through a compromised Safe developer machine, not through a flaw in Safe's smart contracts or source code. Read together, the two documents separate attribution from mechanism: one names who the FBI holds responsible, the other names how the operator's own reviewers say access was gained.

A quorum can be met without a broken signature scheme

Bybit's ether sat behind a multisignature arrangement: several designated signers, each reviewing and approving a transaction before it could move. The signers did not sign in the abstract. They signed what a web interface displayed to them. If that interface can be made to show one transaction while the wallet contract executes another, a signer can produce a technically valid signature over a payload they never intended to authorise. Enough valid signatures still satisfy the multisig threshold; the cryptography performed exactly as designed. What failed was the layer that is supposed to tell a human signer what their signature actually commits to. This is the distinction the Safe Foundation statement draws when it separates a compromised developer machine from a vulnerability in the underlying contracts.

What the record does and does not establish

The FBI notice establishes a date, an amount, and an attribution, and it asks node operators, exchanges and analytics firms to act on a list of associated addresses. It does not describe the technical path into Bybit's signing process. The Safe Foundation statement describes that path but is the operator ecosystem's own account, issued as a remediation update rather than an independent audit result, and it does not say whether every affected signer could have detected the mismatch with tools available at the time. Neither document establishes that multisignature custody generally is less safe than single-key custody; both are consistent with the narrower claim that a signing interface is part of the custody boundary, not a neutral display sitting outside it.

Questions to carry forward

• Can a signer verify the exact contract call a transaction will make, independent of the interface presenting it?
• What would have flagged a mismatch between displayed and executed instructions before signatures were collected?
• Who absorbs a loss when the signature quorum was satisfied but the authorisation behind it was manipulated?

The theft does not show that a threshold of signatures is a weak form of custody. It shows that the threshold only protects what the signers can actually see, and that seeing correctly is a separate engineering problem from signing correctly.