PatchSiren

PatchSiren cyber security CVE debrief

CVE-2026-40320 Giskard-AI CVE debrief

CVE-2026-40320 is a vulnerability in Giskard, an open-source testing framework for AI models. In versions prior to 1.0.2b1, the ConformityCheck class rendered the rule parameter through Jinja2's default Template() constructor, silently interpreting template expressions at runtime. If check definitions are loaded from an untrusted source, a crafted rule string could achieve arbitrary code execution. Exploitation requires write access to a check definition and subsequent execution of the test suite. This issue has been fixed in giskard-checks version 1.0.2b1.

Vendor
Giskard-AI
Product
giskard-oss
CVSS
MEDIUM 5.4
CISA KEV
Not listed in stored evidence
Original CVE published
2026-04-17
Original CVE updated
2026-10-26
Advisory published
2026-04-17
Advisory updated
2026-10-26

Who should care

Defenders responsible for AI model testing and validation should assess exposure to this vulnerability. Specifically, those using Giskard's giskard-checks version prior to 1.0.2b1 should verify their version and consider upgrading to 1.0.2b1 or later.

Why it matters

CVE-2026-40320 is a vulnerability in Giskard's ConformityCheck class that could allow arbitrary code execution if check definitions are loaded from an untrusted source. Defenders should verify their version of giskard-checks and consider upgrading to 1.0.2b1 or later.

  • Potential for arbitrary code execution if check definitions are loaded from an untrusted source
  • Requires write access to a check definition and subsequent execution of the test suite
  • Verification of giskard-checks version is necessary to determine exposure
  • Upgrade to 1.0.2b1 or later may be necessary to mitigate the vulnerability

Technical summary

The ConformityCheck class in Giskard's giskard-checks version prior to 1.0.2b1 renders the rule parameter through Jinja2's default Template() constructor, allowing for silent interpretation of template expressions at runtime. If check definitions are loaded from an untrusted source, a crafted rule string could achieve arbitrary code execution. This requires write access to a check definition and subsequent execution of the test suite. The issue has been fixed in giskard-checks version 1.0.2b1. Defenders should prioritize verifying the version of giskard-checks in use and upgrading to 1.0.2b1 or later if necessary.

Defensive priority

Defenders should prioritize verifying the version of giskard-checks in use and upgrading to 1.0.2b1 or later if necessary.

Recommended defensive actions

  • Verify the version of giskard-checks in use
  • Upgrade to 1.0.2b1 or later if necessary
  • Restrict write access to check definitions
  • Monitor test suite execution
  • Review compensating controls for exposed systems
  • Check relevant monitoring, detection, and logs for exposed assets
  • Track exceptions and retest remediated assets

Evidence notes

The CVE record and NVD entry provide details on the vulnerability, including its description, CVSS score, and affected versions. The vulnerability affects Giskard's giskard-checks version prior to 1.0.2b1. Exploitation requires write access to a check definition and subsequent execution of the test suite. Defenders should verify their version of giskard-checks and consider upgrading to 1.0.2b1 or later. The CVE Program and NVD provide official records and assessments of the vulnerability.

Sources and references

Verified primary and authoritative sources

  • CVE-2026-40320 CVE Program record

    Publisher, destination, and source semantics verified

    URL: https://www.cve.org/CVERecord?id=CVE-2026-40320

    CVE Program - Official CVE Program record with source-provided CVE metadata.

  • CVE-2026-40320 NVD vulnerability detail

    Publisher, destination, and source semantics verified

    URL: https://nvd.nist.gov/vuln/detail/CVE-2026-40320

    NIST National Vulnerability Database - Official NIST NVD detail page and source-specific vulnerability assessment.

Supplemental references

Methodology and review provenance

AI-assisted synthesis based on stored public vulnerability evidence. System validation, approval state, and publication status do not by themselves establish human review of this revision. PatchSiren helps prioritize defensive review and does not prove exposure or remediation on any system.