CCCS-CIC-AndMal-2020unb.ca
The dataset includes 200K benign and 200K malware samples totalling to 400K android apps with 14 prominent malware categories and 191 eminent malware families.
10 datasets
Security datasets are used to train and test detection systems and to track incidents. The list includes CCCS-CIC-AndMal-2020, with 200K benign and 200K malware Android apps, and traffic and log data captured during the Cyber Czech red team and blue team exercise. Open Dependency Risk ranks widely used npm and PyPI packages and is updated daily, and other entries record smartphone security-update end dates, AI agent security incidents, PDF tampering statistics and scam cases.
Malware corpora should be handled in an isolated environment, since the samples are live code. Exercise captures are realistic but come from one event, so models trained on them may not generalise to other networks. Most of the incident and statistics sets here are CC BY or MIT, and many are compiled by a single company, so check how each figure was sourced and when the snapshot was taken.
CCCS-CIC-AndMal-2020unb.ca
The dataset includes 200K benign and 200K malware samples totalling to 400K android apps with 14 prominent malware categories and 191 eminent malware families.
Documented data-breach and scam cases from the Egidio Threat Lab's French corpus, with every figure labelled as either officially confirmed by an authority or merely claimed by an attacker — the two are never aggregated into a single total. Machine-readable JSON, one record per case, with title…
A measurement of GitHub's public bounty board taken from both sides, with every dated scan published so the claims can be re-checked rather than trusted. Collected from public search and issue pages with no login and nothing behind authentication. SIDE ONE, what the board ASKS FOR (2026-08-08). 563…
Open Dependency Riskgithub.com
Daily-updated dataset of 1,500 widely-used npm and PyPI packages ranked by weekly downloads, joined with registry metadata and known vulnerabilities from OSV.dev, plus mechanical risk flags (staleness, deprecation, vulnerability counts). JSON/CSV.
Orca AI Incident ArchiveContinuum AI
17Open database of real-world AI agent security incidents, recorded month by month from January 2025 (365 records as of 2026-09-24, plus one 2024-12 precursor). One Markdown file per event with a YAML header, an attack-chain diagram and at least one clickable primary source; 625 unique source URLs in…
Anonymized aggregate statistics from real-world PDFs analyzed in production by a structural PDF tamper-detection service. Covers the structural modification rate (share of PDFs that reached a confident verdict and showed post-creation structural editing — currently ~48%), the most-common…
Romance Scam Statistics (2024-2026)Arnaques-Rencontres.fr
2Verified, source-attributed statistics on romance scams (confidence fraud), compiled and fact-checked in June 2026. Aggregates figures from the FBI IC3 2025 Internet Crime Report, FTC, City of London Police, France's DGPN, Cybermalveillance.gouv.fr, GASA Global State of Scams and the Chainalysis…
Metadata from a nightly record of the text that MCP (Model Context Protocol) servers, agent skills and plugins publish for AI agents to read (tool descriptions, input schemas, instruction files). One row per entry (93,634), per version with its content hash and tool count (188,088), and per…
Smartphone Software and Security Update SupportDevLifeCheck
Open dataset of smartphone OS-upgrade commitments and security-update end dates for 700+ models across 17 brands, with linked manufacturer sources and per-model DevLifeCheck URLs. CSV/JSON; CC BY 4.0.
This dataset was acquired during Cyber Czech – a hands-on cyber defense exercise (Red Team/Blue Team) held in March 2019 at Masaryk University, Brno, Czech Republic. Network traffic flows and a high variety of event logs were captured in an exercise network deployed in the KYPO Cyber Range Platform.