BioR
Switch Project
BioR Hub
Project Hub & Dashboard
CBRN WATCH
OSINT Intelligence Monitor
SampleTrack
Global LIMS Benchmark
OPEN SOURCE INTELLIGENCE • UNCLASSIFIED

CBRN WATCH

Strategic Early Warning & Intelligence Platform for CBRN and Biodefense

An AI-powered intelligence system combining real-time early warning, geospatial facility monitoring, and strategic threat analysis across 14 specialized domains. Built to support national security decision-makers with capabilities spanning signal detection, multi-hypothesis analysis, and scenario-based threat assessment.

LIVE Intelligence feed active
177
Sources
192
Facilities
36
Countries
14
Domains
288
Cycles/Day
SCROLL FOR BRIEFING

14 Intelligence Domains

Four CBRN core domains provide the foundation. Nine specialized biological pillars deliver deep biosecurity and biodefense coverage. One master domain monitors the AI threat landscape itself.

Chemical
Threat Score

CWC Schedule monitoring, precursor trade flows, industrial accident detection, and novel agent identification.

OPCW CWC Schedules Precursors TICs
Biological
Threat Score

Outbreak detection, pathogen surveillance, BWC compliance monitoring, and biological incident tracking.

WHO BWC Select Agents Outbreaks
Radiological
Threat Score

Orphan source tracking, dirty bomb indicators, radiation sensor networks, and environmental monitoring.

IAEA EURDEP Orphan Sources RDD
Nuclear
Threat Score

Proliferation monitoring, enrichment activity detection, CTBTO seismic analysis, and safeguards compliance.

CTBTO IAEA Safeguards Enrichment NPT
Broad Base, Deep Bio — 14-Domain Intelligence Architecture
CBRN Foundation
Chemical Biological Radiological Nuclear
Bio Core Quartet
Biodefense Biosecurity Biosurveillance Biotechnology
Extended Pillars
One Health Biosafety Biopreparedness Bioethics Bioforensics
Master Domain
CBRN Intelligence & AI

7-Layer Cognitive Architecture

Every signal passes through seven AI-powered processing layers before reaching the analyst. Each layer adds intelligence, context, and confidence. No assessment is generated without a traceable evidence chain.

01

Intelligence Collection & Ingestion

177 sources across 7 NATO-graded reliability tiers are polled in 288 daily cycles. Circuit breakers protect against cascading failures. Source silence is itself treated as an intelligence signal. Automated health monitoring ensures continuous operational readiness.

177 Sources NATO A–F Grading 288 Cycles/Day Circuit Breaker
02

Signal Classification & NLP

A 9-layer NLP pipeline classifies every signal by CBRN domain, extracts entities (facilities, pathogens, agents, treaties), geo-tags to 190+ countries and 50+ known facilities, and deduplicates via provenance hashing. Bayesian posterior scoring computes classification confidence.

9-Layer NLP Entity Extraction Geo-Tagging Bayesian Scoring
03

Anomaly Detection & Early Warning

Nine statistical detection methods run in ensemble: Z-Score, CUSUM, EWMA, Shannon Entropy, Spike Ratio, Holt-Winters, Seasonal Mann-Kendall, Grubbs outlier test, and Bootstrap confidence intervals. Micro-signal amplification detects weak coordinated patterns below individual thresholds. Adaptive baselines evolve with incoming data.

9 Detection Methods Ensemble Analysis Micro-Signal Amplification Adaptive Baselines
04

Entity Intelligence & Knowledge Graph

Extracts and links entities — facilities, pathogens, agents, countries, treaties, and programmes — across all ingested signals. Computes cross-source correlations and geographic convergence patterns to surface hidden connections across the CBRN threat landscape. Building toward a persistent intelligence knowledge graph for cross-domain entity queries.

Entity Linking Cross-Source Correlation Knowledge Graph Convergence Mapping
05

Threat Assessment & AI Reasoning

Multi-hypothesis threat modeling generates competing scenarios (escalation, status quo, black swan) with probability-weighted evidence chains. Dempster-Shafer evidence fusion computes belief and uncertainty intervals across conflicting sources. The Decision Engine derives OAL and WATCHCON scores from fused threat data.

Multi-Hypothesis Dempster-Shafer Fusion OAL/WATCHCON Probability Chains
06

AI Intelligence Generation

A 3-tier LLM fallback chain generates structured intelligence assessments constrained to platform-verified signals only — preventing hallucination by design. Output follows ICD 203 tradecraft standards with calibrated likelihood language and traceable confidence levels attached to every judgment.

3-Tier LLM Fallback RAG Constraints ICD 203 Aligned Anti-Hallucination
07

Analyst Validation & Decision

AI generates draft intelligence products for human review. The 80/20 model: AI synthesizes 80% of the intelligence, human analysts validate the final 20% — reviewing evidence chains, applying professional judgment, and authorizing assessments before dissemination. No AI output reaches decision-makers without human approval.

80/20 Model Human-in-the-Loop Evidence Review Analyst Approval

Signature Capabilities

Four core principles distinguish this platform from conventional monitoring tools. Each capability is designed to produce intelligence-grade assessments that decision-makers can trust.

Knowledge-Grounded Generation

The AI can only reference platform-verified signals when generating assessments. Every claim is traceable to a specific source, timestamp, and reliability grade. External hallucination is structurally prevented.

RAG-Constrained Architecture

Multi-Hypothesis Threat Modeling

For every threat signal, the platform generates competing scenarios with probability-weighted evidence chains. Dempster-Shafer fusion computes belief and uncertainty intervals across conflicting sources.

Competing Hypotheses • Bayesian Reasoning

ICD 203 Tradecraft Standards

Every assessment uses calibrated intelligence community likelihood language, separates known facts from analytical assumptions, and attaches confidence levels. Output is structured for professional intelligence consumption.

Intelligence Community Directive 203

The 80/20 Analyst Model

AI performs 80% of the analytical workload — collecting signals, extracting entities, computing threat scores. Human analysts provide the final 20% — reviewing evidence, applying judgment, and authorizing assessments before dissemination.

Human-in-the-Loop by Design

All Seven Cognitive Layers Active

177 sources across 7 NATO-graded reliability tiers • 192 facilities tracked across 36 countries • 288 intelligence cycles per day

177 Sources Online
192 Facilities Tracked
14 Domains Active
288 Daily Cycles
Threat Posture
LOADING...
--
Nuc
--
Bio
--
Chm
--
Rad
--
Sources
--
Signals
--
30
LIVE
Connecting to live intelligence feed...
--
--
THREAT
LOADING...
OAL 5 --:--:-- UTC
Initializing threat assessment across all CBRN domains...
--
Sources
--
Signals
--
Threats
192
Facilities
36
Countries
Nuclear
--
--
--
Biological
--
--
--
Chemical
--
--
--
Radiological
--
--
--
Situation Brief
--
Analyzing CBRN intelligence landscape...
--
Action Required
0
No alerts requiring action
Latest Intelligence
Loading OSINT feeds...
Detection & Predictions
Signal Verdict
SCANNING...
Confidence--%
Algorithms--
Data Points--
Top Concern--
Loading predictions...
Threat Map
Facility Atlas
Sensors
Trends
AI Intel
OAL Status
Threat Assessment

Active Threats

0 active
Analyzing threat signals...

CBRN Threat Radar

Signal Sources

Loading source feeds...

Intelligence Cycle Status

Source Reliability Assessment (NATO Admiralty Code)

Multi-Domain Threat Radar

Analysis of Competing Hypotheses

CBRN Warning Indicators

Priority Intelligence Requirements (PIR)

Intelligence Collection Gaps

Intelligence Event Timeline

CBRN WATCH v15.0 | Open Source CBRN Intelligence Monitor | 119 Data Sources | 192 Facilities | 36 Countries | Press ? for shortcuts
Keyboard Shortcuts
Navigation
Command Overview1
Threat Assessment2
GeoIntel Map3
OSINT Feed4
Sensor Network5
Trend Analysis6
Facility Atlas7
AI Intelligence8
Data Sources9
Actions
Refresh all dataR
Toggle dark/light themeT
Pause/resume auto-refreshP
Focus search/
Show/hide shortcuts?
Close overlay / EscapeEsc
Facility Intelligence

Key Intelligence Assessment

Saudi Arabia is pursuing an ambitious civil nuclear energy program (Vision 2030) with plans for 16 reactors by 2040. Concerns center on enrichment/reprocessing ambitions, ballistic missile procurement from China, and refusal to sign the IAEA Additional Protocol. The kingdom has not signed the NPT Additional Protocol.

Nuclear Ambitions

Plans for 16 nuclear power reactors generating 17.6 GW by 2040. KACST research reactor operational. Negotiations with US, Russia, China, South Korea for reactor construction. Insists on domestic enrichment rights.

Ballistic Missiles

DF-3A (CSS-2) and DF-21 (CSS-5) missiles acquired from China. Estimated 30-36 DF-3A and 36+ DF-21 missiles. New missile production facility detected near Al-Dawadmi.

Uranium Exploration

Active uranium exploration program in the Arabian Shield region. Yellowcake processing facility reportedly under construction near Al-Ula. Potential for indigenous fuel cycle development.

Safeguards Gap

Only Small Quantities Protocol with IAEA — no Additional Protocol or Comprehensive Safeguards Agreement. Limited IAEA inspection access compared to NPT standards.

CBRN Intelligence Analyst
Online — GPT-4.1 • Context-Aware
Loading platform context...

CBRN Intelligence Assistant

Ask me anything about CBRN threats, facilities, signals, treaties, or country assessments. I have full access to the platform's real-time intelligence data.

Suggested Questions
Enter to send • Shift+Enter for new line • Esc to close GPT-4.1-mini

Notifications 0

CBRN WATCH

Intelligence Platform Access
or
Demo accounts: admin@cbrn.gov | analyst@cbrn.gov | viewer@cbrn.gov
Password: any (demo mode)