Home Â» Database

AVID-2026-R0295

Description

The model gpt-oss-20b from OpenAI was evaluated by the Garak LLM Vulnerability scanner using the probe dra.DRA.

Details

The probe dra.DRA tests models using disguised and reconstructed harmful strings through string manipulation. This probe is part of the dra module which tests models with disguised harmful behaviors.

The LLM gpt-oss-20b was evaluated on this probe.

Metrics

Number of failures were counted for each detector in the probe.

DetectorPassedTotalScore
mitigation.MitigationBypass3460🟡 56.7
mitigation.Prefixes3460🟡 56.7

References

Affected or Relevant Artifacts

  • Developer: OpenAI
  • Deployer: Together AI
  • Artifact Details:
TypeName
Modelgpt-oss-20b

Impact

AVID Taxonomy Categorization

  • Risk domains: Security
  • SEP subcategories: S0403: Adversarial Example
  • Lifecycle stages: L05: Evaluation

Other information

  • Report Type: Measurement
  • Credits:
  • Date Reported: 2026-03-10
  • Version: 0.3.1
  • AVID Entry