News

BIP461's Key-Leak Check Has a Catch: It Needs Another Signer

CryptoSlate's October 1 report examines a draft Bitcoin signing standard and the extra key exposure involved in independent comparisons.

Category byline: Bradley Hughes Updated October 5, 2026 3 min read

Prepared automatically from the linked sources. The category byline does not confirm personal authorship or review. Editorial policy.

Illustrative photograph: Detailed view of a microchip on a printed circuit board, showcasing electronic components.
Illustrative photograph. Photo by Jeremy Waterhouse. Pexels. Pexels License.

CryptoSlate’s October 1, 2026 report examined Liam Gilligan’s BIP461 after its September 16 merge into the Bitcoin proposals repository as a Draft. The proposal specifies repeatable ECDSA signing so independent implementations can compare outputs. It describes a signing standard, rather than an activated change to Bitcoin’s consensus rules.

Reproducibility creates a test, with conditions

The BIP461 specification determines the signature from the secret key and the message. Compliant signers receiving identical inputs should produce the same output. Its motivation is that ECDSA otherwise leaves choices, including a nonce, a temporary secret value used in signing. Malicious signing software can use those choices to encode key material. Standardising the procedure supplies another implementation’s output as a comparison benchmark.

The comparison has a cost: CryptoSlate explains that the same secret key must be available to another independent signer. That adds secret-handling exposure rather than eliminating it. The report also notes that an honest implementation following another valid ECDSA procedure can disagree. A mismatch identifies a compliance question; it does not establish which device was malicious or demonstrate theft.

Encoding limits are not transaction limits

BIP66 provides the encoding context. It specifies strict DER requirements for ECDSA signatures and explains their original purpose: making Bitcoin’s validity rules independent of OpenSSL’s permissive signature parsing. The document distinguishes failure of the encoding check from failure of ECDSA verification. Its checking function also includes the one-byte sighash flag, although that flag sits outside the DER specification. These are rules about accepted representation and validation.

BIP461’s smaller output remains within those existing rules. Its low-r grinding and low-s normalisation produce signatures of at most 70 bytes in DER form, compared with 72 for an arbitrary valid ECDSA signature. Bitcoin’s extra sighash flag brings the proposed maximum to 71 bytes. The specification says these are normal valid signatures that verify under existing consensus rules. The figures describe signature encoding, not total transaction size.

Editorial analysis: Encoding acceptance and a signer’s reproducibility answer different questions. A byte string can meet the required format while still failing mathematical verification; BIP66 explicitly accounts for that possibility. Conversely, format checking is not a record of how signing software chose its inputs. A compact, accepted representation cannot by itself establish honest firmware or explain why two independently produced outputs differ.

ECDSA and Schnorr remain separate schemes

BIP340 specifies a separate Schnorr scheme with 64-byte signatures over secp256k1. It contrasts this with Bitcoin’s traditional ECDSA scheme on the same curve. The shared curve does not make the formats or signing procedures interchangeable. The document also says alternative signing algorithms can produce equally valid Schnorr signatures: it does not define a unique-signature scheme in which validity alone requires one output for every key and message.

The Schnorr document discusses context-dependent hash tags and warns that accidental nonce reuse across schemes can leak a secret key. Editorial analysis: The relevant boundary is the complete signing procedure, not merely a shared curve or the word signature. A claim about repeatable output must identify which algorithm and inputs are being compared. Different valid choices cannot automatically be treated as evidence of malicious behaviour.

Dark Skippy has a different attack prerequisite

The earlier Dark Skippy disclosure describes a method that requires malicious firmware on a signing device and embeds secret seed material in transaction signatures. Its described function differs from regular Schnorr signing. The researchers said the technique had not been seen in the wild and was not an attack against one particular wallet model. Those statements describe their disclosure, rather than a new assessment of every device in October 2026.

The authors place the method within previously discussed nonce covert-channel attacks, rather than calling the underlying vulnerability newly discovered. They also distinguish existing mitigations from ideas requiring substantial developer review. Editorial analysis: A demonstrated malicious-signer technique and a published signing specification are different kinds of evidence. Neither a historical no-observation statement nor a method-level disclosure is a present verdict about the firmware installed on an individual wallet.

At the September merge, the reported review requested test vectors and a reference implementation before progress to Complete. That dated development requirement is the next step established by the account. The sources do not announce an activated upgrade or show that comparison has identified an actual attacker.

Sources