Evidence boundary. The cited research supports analysis of production Android Automotive OS firmware from four vehicle manufacturers, SBOM reconstruction, CVE mapping and reproduction of CVE-2023-5217 on an affected AAOS 10 image. Escrow, triggered maintenance rights and lifecycle sovereignty are engineering recommendations, not explicit requirements of UN R156 or ISO/SAE 21434.

Authenticity and vulnerability are different properties

Research published in Computers & Security analysed production Android Automotive OS firmware from four vehicle manufacturers, reconstructed software bills of materials from the binaries and mapped embedded components to known vulnerabilities. The researchers also reproduced CVE-2023-5217 through a WebView-based test on an AAOS 10 image that contained the vulnerable library version.

The important point is not that firmware signing failed. The images were legitimate. The security gap was vulnerability retention inside software that could still pass the authenticity checks expected of an approved image.

A signature answers a narrow but essential question: was this artefact signed by an authorised key and has it changed since signing? It does not prove that every dependency is current, that the build contained no vulnerable component, or that the release remains defensible months or years later.

What this diagram shows

Firmware trust has several layers. Cryptographic authenticity is necessary, but vulnerability status and lifecycle support remain separate evidence domains.

Firmware trust chainA valid signature authenticates the release authority, not the security state of every component inside the image
Residual security risk Governed evidence domain
Release and lifecycle evidence
Release authoritySigning keyProves authorised cryptographic origin
Artefact integritySigned firmwareDetects unauthorised modification after signing
Residual riskEmbedded dependencyA vulnerable component can remain inside an authentic image
Lifecycle authorityPatch and release capabilityThe organisation must still be able to fix, validate and deploy
Decision gate

Can the organisation prove both that the firmware is authentic and that its vulnerable dependencies can still be remediated through the supported lifecycle?

YESThe signature is one control inside a defensible release process.
NOCryptographic trust remains valid while lifecycle risk becomes unmanaged.

The difficult part starts after the CVE is found

Replacing a vulnerable component in a production vehicle can affect interfaces, timing, memory, dependencies and regression evidence. UN Regulation No. 156 requires manufacturers to manage software updates and assess their impact, while the exact approval consequences depend on the functions and vehicle characteristics affected by the change.

That creates a supply-chain decision problem. The OEM wants the exposure reduced. A Tier 1 may point to an accepted baseline. The software supplier may provide an upstream fix without carrying the cost of vehicle-level integration and validation. A temporary functional restriction can reduce attack surface while creating warranty, customer or contractual consequences.

ISO/SAE 21434 requires cybersecurity activities and responsibilities to be managed across the supply chain. The practical interface should therefore define who investigates exploitability, who implements the fix, who validates it, who funds temporary mitigation, and who owns external communication when evidence is incomplete.

The lifecycle can outlive the supplier

The harder failure mode appears when the original supplier no longer has the capability or commercial incentive to maintain the dependency. A vehicle can remain in service long after a proprietary component changes owner, loses its original engineering team or becomes unsupported.

For critical dependencies, continuity may require more than source-code access. Reproducible build environments, toolchain availability, dependency records, signing and release procedures, triggered maintenance rights and knowledge transfer may be necessary to preserve the practical ability to produce a trustworthy update.

ISO 24089 reinforces the need to manage software-update responsibilities across manufacturers and suppliers. The engineering objective is therefore not simply to preserve code. It is to preserve the complete capability to build, validate, authorise and release the next version safely.

A supplier can transfer software. It cannot transfer lifecycle risk.
The decision
Treat firmware signing as one evidence layer, and contract for the practical capability to remediate vulnerable software across the full vehicle lifecycle.
Operational checks
  • Map signed firmware to an SBOM and vulnerability status for the exact production variant.
  • Separate cryptographic authenticity evidence from exploitability and patchability evidence.
  • Define supplier responsibility for investigation, correction, validation and temporary mitigation.
  • Identify critical proprietary dependencies whose support horizon is shorter than the product lifecycle.
  • Preserve the build, toolchain, signing and release capability needed to maintain those dependencies if the supplier relationship ends.
Where to go next

Continue this decision.

Recommended next · Product Security · Supply Chain · AutomotiveA VEX Statement Is a Claim That Needs Evidence

VEX can suppress false urgency, but an unsupported “not affected” status simply converts vulnerability noise into governance risk.

Source record

Sources & further reading

5 cited sourcesHow we source →
← All analysisCompanion episode →