▸ METHODOLOGY

How AntiBrute works

AntiBrute aggregates open cyber-threat intelligence and renders it as a live, AI-narrated map of global activity. No proprietary feeds — all sources are public and cited.

DATA SOURCES
  • NVD● LIVE · 30 min

    National Vulnerability Database — official US CVE feed. Every CVE on the site links back to its NVD record.

  • CISA KEV● LIVE · 60 min

    Known Exploited Vulnerabilities catalog — CVEs with confirmed active exploitation in the wild.

  • GitHub Security Advisories● LIVE · 60 min

    GHSA database — vulnerabilities affecting open-source packages across npm, PyPI, Maven, RubyGems, Go, Composer, NuGet, and more.

  • AbuseIPDB● LIVE · 6 h

    Crowdsourced blacklist of IP addresses actively scanning, brute-forcing, or attacking systems. Powers the Top Threat Actors panel and country attribution.

  • HIBP○ PLANNED · planned

    Have I Been Pwned — public breach disclosures.

  • GDELT○ PLANNED · planned

    Global event database — geopolitical & news context.

▸ EVERY INCIDENT STORES ITS SOURCE URL AND RETRIEVAL TIMESTAMP. CLICK ANY [SOURCE] BADGE TO READ THE ORIGINAL ADVISORY.

AI LAYER

A multi-model AI summarization layer (frontier LLMs from leading providers, accessed through a hardened internal gateway) converts raw advisories and breach disclosures into three parallel narratives:

  • ▸ PRO — preserves CVSS, IOCs, and TTPs as-is.
  • ▸ PLAIN — rewrites into jargon-free English.
  • ▸ EXEC — extracts business impact and recommended action.

Per-country narratives and the daily brief are regenerated on a schedule and cached. AI is never used for severity scoring or threat attribution — those come directly from upstream sources.

RISK SCORING

Country and industry risk scores are weighted composites of:

  • Active critical CVEs touching deployed tech in the geography
  • Observed ransomware / APT events in the trailing 30 days
  • Botnet activity and abuse reports per capita
  • Geopolitical context signals (GDELT)

Output range: 0–100. Weights and formulas are documented in the public methodology repo (link forthcoming).

LIMITS & DISCLAIMERS
  • ▸ AntiBrute is informational. Not security advice.
  • ▸ Coverage skews to incidents that are publicly disclosed.
  • ▸ Country attribution reflects observed activity, not absolute origin.
  • ▸ AI summaries can be wrong. Always check source advisories before acting.
▸ ANTIBRUTE.COM // BUILT FOR DEFENDERS, EXECS, AND THE CURIOUS PUBLIC