LLM-Generated Phishing and Legitimate Emails for Detection Research

A collection of phishing and legitimate emails generated using Large Language Models, specifically DeepSeek for Chinese emails and OpenAI models for English emails. The dataset is intended to facilitate research and development in phishing email detection and classification. It was created by Dizzzy0x00 and last updated on December 13, 2025.

Use Cases

• Train phishing email classifiers based on LLM-generated email content.
• Benchmark detection algorithms against synthetic phishing attempts.
• Study linguistic patterns distinguishing phishing from legitimate emails.
• Analyze cross-lingual phishing tactics in Chinese and English contexts.

Strengths

• Emails are generated using specific, named LLMs (DeepSeek and OpenAI models).
• Dataset includes content in two languages: Chinese and English.
• Explicitly designed for phishing detection and classification research.

Limitations

• Row count and dataset size are unknown, limiting suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

huggingface

Collection Method

Generated using Large Language Models (LLMs).

Freshness

Last updated 2025-12-13 06:09:21; freshness should be verified.