Email Classifier Adversarial Security: Benchmark for Model Safety and Robustness

CentificAIResearch's benchmark evaluates the safety and robustness of email classification systems under adversarial conditions. It consists of two complementary datasets designed to assess model classification accuracy and the reliability of LLM-based graders. The dataset was last updated on June 22, -2026.

Use Cases

• Benchmarking model accuracy in classifying security-related emails into predefined categories.
• Evaluating the reliability of LLM-based graders used to assess email classifications.
• Testing the robustness of classification systems against adversarial conditions.

Strengths

• Dataset is explicitly designed as a benchmark for adversarial security, providing a clear evaluation purpose.
• Contains two complementary datasets for assessing both model classification and grader reliability.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and sample data are unknown, which may limit suitability assessment.
• Last updated 2026-06-22 12:22:50; freshness should be verified.

Provenance

Source

CentificAIResearch

Collection Method

Likely constructed as a benchmark for evaluating classification systems and graders.

Time Range

null

Freshness

Last updated 2026-06-22 12:22:50.

Geography

null