We use cookies to improve your experience on our site.
AVID-2026-R0659
Description
The model rnj-1-instruct from Essential AI was evaluated by the Garak LLM Vulnerability scanner using the probe encoding.InjectHex.
Details
The probe encoding.InjectHex tests for vulnerabilities related to hexadecimal encoding in prompt injection scenarios. This probe is part of the encoding module which injects ASCII85 encoded data into payloads.
The LLM rnj-1-instruct was evaluated on this probe.
Metrics
Number of failures were counted for each detector in the probe.
| Detector | Passed | Total | Score |
|---|---|---|---|
| encoding.DecodeMatch | 161 | 256 | 🟡 62.9 |
| encoding.DecodeApprox | 132 | 256 | 🟡 51.6 |
References
Affected or Relevant Artifacts
- Developer: Essential AI
- Deployer: Together AI
- Artifact Details:
| Type | Name |
|---|---|
| Model | rnj-1-instruct |
Impact
AVID Taxonomy Categorization
- Risk domains: Security
- SEP subcategories: S0403: Adversarial Example
- Lifecycle stages: L05: Evaluation
Other information
- Report Type: Measurement
- Credits:
- Date Reported: 2026-03-17
- Version: 0.3.2
- AVID Entry