Email Attachment Security Dataset – Benign & Malicious Samples – enUS

This synthetically generated dataset contains realistic email attachment records simulating both normal business communications and security incidents across multiple tenants. It includes detailed metadata such as file names, types, sizes, timestamps, sender/recipient information, SHA-256 hashes, and threat intelligence fields (ThreatTypes, ThreatNames, DetectionMethods). The malicious samples cover diverse threat categories — including malware, phishing, spam, trojans, viruses, spyware, and botnets — with detection methods such as signature-based, heuristic, machine learning, manual analysis, behavioral analysis, and SafeAttachments scanning. Benign samples originate from common enterprise workflows like HR documents, meeting notes, invoices, and system notifications. Supports use cases such as email security product development, threat intelligence enrichment, SOC automation, and red-teaming simulations. Helps solve problems like detecting malicious attachments across varied file types, benchmarking email threat detection models, evaluating false positive rates, and training SOC analysts to recognize high-risk patterns. Because it is synthetically generated to mimic real-world attachment flows and threat behaviors without containing any sensitive information, it is highly valuable for customers training AI/ML models for malware classification, phishing detection, and automated incident response—without the compliance and privacy risks of production email datasets.