A Bitcoin hardware wallet can display the correct payment and produce a mathematically valid signature while secretly revealing information about its owner’s private key. On November 13, 2026, Bitcoin Optech highlighted two new draft proposals from Connor Avila intended to close a dangerous communication channel between compromised signing devices and outside observers. The proposed protection is called anti-exfiltration signing.
A hardware wallet normally keeps the private key inside the device. Yet the device also selects a temporary secret number, or nonce, when signing. A deliberately manipulated nonce can encode information in signatures that still pass Bitcoin’s ordinary validity checks. The Dark Skippy research demonstrated how a malicious signing device could exploit the technique. Such an attack requires a compromised or malicious signer; it is not evidence that every hardware wallet leaks secrets.
Video context: Stephan Livera’s August 2024 Dark Skippy discussion explains the older attack and possible mitigations, not a demonstration of Avila’s November 2026 draft.
Why a valid Bitcoin signature is not enough
Bitcoin checks whether a signature satisfies its cryptographic rules, not whether the signer chose its nonce to hide a message. A malware-infected signing device might deliberately vary otherwise valid signatures. Repeating a nonce can also directly expose an ECDSA key, although nonce reuse and covert exfiltration are distinct failure modes. BitcoinVersus.Tech recently covered research into recoverable keys from repeated ECDSA nonces. The new drafts address a different threat: intentional leakage by the device entrusted with signing.
How the two-round defense works
Under Avila’s proposed design, a host computer first commits to fresh random data without revealing the data itself. The signing device responds with a commitment to its initial nonce point. Only afterward does the host reveal its randomness. The signer combines the contributions to produce a signature, and the host checks that the committed nonce and final result match the protocol.
Order matters. A malicious signer cannot inspect the host’s secret contribution before committing to its initial point, reducing its freedom to encode chosen information. The design covers both ECDSA and BIP340 Schnorr signatures. A companion draft adds three proposed per-input fields to Partially Signed Bitcoin Transactions (PSBTs) so compatible wallets and signing devices can exchange the required commitments.
Video context: Bitcoin University’s September 2024 explanation provides additional archival background on Dark Skippy. Neither video reports a deployed 2026 anti-exfil standard.
Important security limits
The drafts are proposals, not activated Bitcoin consensus changes. Avila’s specification includes test vectors, a Python reference implementation and C integration work, but the author explicitly says the cryptographic arguments have not yet received an independent audit. An implementation must also keep the host’s random value confidential from third parties. Air-gapped transfer methods are not automatically safe under the proposed assumptions.
Failed signing attempts matter because a malicious device could leak information by selectively refusing to sign. The draft therefore proposes persistent failure tracking and user warnings. It does not solve compromised key generation, physical extraction, malicious host-and-device collusion, or every multisignature scenario. Backward-compatible support will depend on wallet and hardware manufacturers implementing and reviewing interoperable software.
What Bitcoin holders should watch next
The immediate development milestone is code review, independent cryptographic scrutiny and eventual interoperable wallet support, not a forced wallet migration. Holders should continue verifying transaction details on their signing devices and obtain firmware from trusted sources. For complementary security coverage, see our reports on hardware-wallet backup design, hardware tampering investigations, and experimental post-quantum signing research. Each addresses a separate security problem.
Primary sources and verification
- Bitcoin Optech Newsletter #431, November 13, 2026, draft-proposal overview.
- Avila’s anti-exfil signing specification, draft covering Schnorr and ECDSA.
- PSBT companion specification, draft transport fields and restrictions.
- Bitcoin Optech #431 recap, November 17, 2026, related technical discussion.
Editorial note: The featured cover is an original conceptual anime illustration, not a photograph of a specific device, engineer or product. Neither the security proposal nor the embedded videos have been represented as independently tested by BitcoinVersus.Tech.
We volunteer daily to ensure the credibility of the information on this platform is Verifiably True. If you would like to support to help further secure the integrity of our research initiative, please scan the QR code below.


Leave a Reply