Ledger's 2026 disclosures cover two different failures in the path between what a user reviews and what the device ultimately signs. LSB-023 describes a command-interleaving condition where displayed and signed parameters could diverge. Ledger says application-level guards shipped in Ethereum app 1.22.2 on August 13 and the SDK-level correction followed in…
Inspect the ClaimsCase · DiggingBeagle record
Ledger fixed Ethereum signing-flow flaws found during internal automated/AI-assisted review
Ledger disclosed multiple Ethereum-app signing-flow vulnerabilities found through internal continuous and automated security review. One command-interleaving bug could desynchronize what the user reviewed from what was signed; another swap-path flaw could substitute a token approval for the intended payment. Ledger reported no known exploitation against users and released application/SDK fixes in August 2026.
Scope
Two separate Ledger Ethereum signing-flow vulnerability classes discovered during internal continuous/automated review. Ledger's Cerberus material describes AI-assisted security research, but the public bulletins do not attribute every discovery step to autonomous agents.
At a glance
Hardware-wallet security depends on a cryptographic binding between user-reviewed intent and the exact operation that will be signed. A secure display is not enough if protocol state can be interleaved or an operation type can be substituted. The case also shows a useful AI-security distinction: AI may improve defensive discovery without being the actor…
Read the implicationsPublic materials do not provide a step-by-step mapping of which Cerberus agent or human reviewer discovered each Ledger-internal flaw.
Limits and uncertaintyFull account
Ledger's 2026 disclosures cover two different failures in the path between what a user reviews and what the device ultimately signs. LSB-023 describes a command-interleaving condition where displayed and signed parameters could diverge. Ledger says application-level guards shipped in Ethereum app 1.22.2 on August 13 and the SDK-level correction followed in Secure SDK v26.6.1 on August 21.
LSB-025 describes a separate swap-flow problem: a token-approval operation could be accepted in place of the intended payment. The attack was constrained by the beneficiary and quantity already accepted for the swap, and moving the approved funds still required a later transaction. Ledger says the missing operation-type check was identified on April 28 and fixed in Ethereum app 1.22.3 on August 25.
The AI role belongs on the discovery side. Ledger's Cerberus material describes a six-agent research harness for iterative finding, proof and patching, but the bulletins characterize the work as internal continuous or automated review and do not provide a per-agent attribution for these specific findings. Ledger reports no known user exploitation.
Mechanism and trust boundary
- 01
Command stream and review state can diverge
LSB-023 describes interleaving that can desynchronize what the user reviewed from the parameters ultimately signed.
Boundary: trusted display / signed command state
- 02
Swap flow can confuse operation type
LSB-025 describes accepting a token approval where the intended operation was a payment.
Boundary: user-approved swap intent / transaction operation type
- 03
Existing parameters constrain the swap flaw
The beneficiary and quantity remain tied to already accepted swap parameters, and a separate later transaction is needed to move funds.
Boundary: approval authority / later asset transfer
- 04
Internal automated review finds and fixes the defects
Ledger patches the app and SDK while reporting no known exploitation against users.
Boundary: research finding / production remediation
Timeline
- Apr 28, 2026discovery
Swap-flow operation-type issue identified
Ledger says internal automated review identified the missing operation-type check later documented in LSB-025.
- Aug 13, 2026mitigation
Ethereum app 1.22.2 adds application-level guards
Ledger reports application-level protection for the command-interleaving class.
- Aug 21, 2026mitigation
Secure SDK v26.6.1 addresses the SDK layer
Ledger reports the SDK-level correction for the command-interleaving issue.
- Aug 25, 2026mitigation
Ethereum app 1.22.3 fixes the swap-flow issue
Ledger publishes the application fix for LSB-025.
Claims & evidence
CLM-AILedger's public Cerberus material describes a six-agent AI security harness used for iterative vulnerability discovery, proof and patching, but the security bulletins themselves characterize these findings as internal continuous/automated review rather than attributing every step to autonomous AI.supported
Basis: reported finding
- supportsLedger - Meet Cerberus AI security harnessrole not specified
SRC-LEDGER-CERBERUS-SEP04
Ledger's public Cerberus material describes a six-agent AI security harness used for iterative vulnerability discovery, proof and patching, but the security bulletins themselves characterize these findings as internal continuous/automated review rather than attributing every step to autonomous AI.
CLM-SWAPLedger LSB-025 documents a separate swap-flow issue in which a token approval could be accepted in place of the intended payment.supported
Basis: reported finding
- supportsLedger Security Bulletin 025role not specified
SRC-LEDGER-LSB025
Ledger LSB-025 documents a separate swap-flow issue in which a token approval could be accepted in place of the intended payment.
CLM-IMPACTFor the swap flaw, exploitation required control of the transaction submitted during a swap; the beneficiary and quantity remained constrained to the already accepted swap parameters, and a separate later transaction was required to move funds.supported
Basis: reported finding
- supportsLedger Security Bulletin 025role not specified
SRC-LEDGER-LSB025
For the swap flaw, exploitation required control of the transaction submitted during a swap; the beneficiary and quantity remained constrained to the already accepted swap parameters, and a separate later transaction was required to move funds.
CLM-SWAPFIXThe swap-path approval/payment confusion was fixed in Ethereum app 1.22.3, published August 25, after internal automated review identified the missing operation-type check on April 28.supported
Basis: reported finding
- supportsLedger Security Bulletin 025role not specified
SRC-LEDGER-LSB025
The swap-path approval/payment confusion was fixed in Ethereum app 1.22.3, published August 25, after internal automated review identified the missing operation-type check on April 28.
CLM-TIMELINELedger says application-level guards for the command-interleaving class appeared in Ethereum app 1.22.2 on August 13, with SDK-level correction in Secure SDK v26.6.1 on August 21.supported
Basis: reported finding
- supportsLedger Security Bulletin 023role not specified
SRC-LEDGER-LSB023
Ledger says application-level guards for the command-interleaving class appeared in Ethereum app 1.22.2 on August 13, with SDK-level correction in Secure SDK v26.6.1 on August 21.
CLM-NOEXPLOITLedger reports no known exploitation against users for the disclosed command-interleaving issue.contested
Basis: reported finding
- supportsLedger Security Bulletin 023role not specified
SRC-LEDGER-LSB023
Ledger reports no known exploitation against users for the disclosed command-interleaving issue.
CLM-INTERLEAVELedger LSB-023 documents a command-interleaving condition where displayed and ultimately signed parameters could diverge.supported
Basis: reported finding
- supportsLedger Security Bulletin 023role not specified
SRC-LEDGER-LSB023
Ledger LSB-023 documents a command-interleaving condition where displayed and ultimately signed parameters could diverge.
CLM-NOEXPLOIT2Ledger states it has no evidence that the disclosed Ethereum signing issues were exploited against users.supported
Basis: reported finding
- supportsLedger Security Bulletin 025role not specified
SRC-LEDGER-LSB025
Ledger states it has no evidence that the disclosed Ethereum signing issues were exploited against users.
Implications
Hardware-wallet security depends on a cryptographic binding between user-reviewed intent and the exact operation that will be signed. A secure display is not enough if protocol state can be interleaved or an operation type can be substituted. The case also shows a useful AI-security distinction: AI may improve defensive discovery without being the actor that created or exploited the vulnerability.
Controls and mitigations
- Update the Ledger Ethereum app to versions containing the published fixes.
- Use Secure SDK versions containing the command-state correction.
- Bind displayed intent, operation type and signed parameters in one verifiable state transition.
- Keep per-finding human and automated discovery attribution explicit when reporting AI-assisted security research.
Unknowns and contradictions
- Public materials do not provide a step-by-step mapping of which Cerberus agent or human reviewer discovered each Ledger-internal flaw.
Ledger reports no known exploitation against users for the disclosed command-interleaving issue. has a contested assessment or contradictory evidence.
Sources and citation
Material revision history
- Sep 18, 2026 · Published version · first publication · revision 4
- Sep 19, 2026 · Published version · material revision · revision 17Public fields changed: attack_path, implications, mitigations, narrative, scope, timeline