Autonomous AI Cyberattack Propagation Map & Defense Simulator

Threat Parameters & Defenses

Primary Entry Vector

Propagation Velocity 1.8x

Stealth / Evasion Factor 0.75

Agent Replication Ceiling 12

Autonomous Defense Countermeasures

Micro-Segmentation Dynamic VLAN isolation upon breach

AI Security Counter-Swarm Autonomous bot-vs-bot purges

Honeypot Decoy Traps Synthetic nodes absorbing AI agents

Cryptographic Air-Gap Hard disconnect core grid subnets

Selected Infrastructure Node

Select a node on grid

Click any node to inspect kill-chain status, inject initial infection, or trigger localized air-gap.

Topology Sandbox: Taiwan Critical Infrastructure Grid Cycle: 0s

Secure Subnet

Targeted / Recon

Compromised SCADA

Defense Agent Active

Attack Velocity vs. Containment Rate Analytics Real-time Telemetry Stream

Telemetry & Incident Log

Breached Nodes 0 / 16

Velocity Rate 0.0 n/s

Swarm Purges 0

Grid Integrity 100%

AI Agent Kill-Chain Phase

Phase: Standby

Real-Time Threat Log

[00:00] System initialized. Topology ready. Select entry point or start simulation.

Defensive propagation fixture and integrity captions

Your browser does not support video.

Read the explanation

The propagation simulator bundles sixteen topology nodes and fifteen edges. Two nodes begin with defense status rather than secure status. Its metrics nevertheless compute secure as sixteen minus breached minus targeted, including those defense nodes in the caption. Bars span one hundred ninety two and one hundred eighty at twelve pixels per node and edge record. These are fictional defensive exercise inputs with geographic labels, not a validated real infrastructure map or observations of an actual compromise. Node weights, propagation parameters and containment switches are illustrative assumptions. No operational access to the named infrastructure or actual attack traffic occurs in this diagnostic. Integrity is a deterministic caption: start at one hundred and subtract four times every breached node weight, constrained to a minimum zero. One hypothetical weight-eight node lowers it to sixty eight; one weight-ten node lowers it to sixty. Their bars span one hundred thirty six and one hundred twenty at two pixels per remaining point. These numbers are not electrical grid capacity, monetary loss or measured service availability. The velocity caption divides current breached-node count by the greater of one and cycle number, despite being labeled nodes per second. Counts and simulation cycles are different from elapsed wall-clock time, so the label is not a measured throughput statistic. An illustrative propagation branch derives a raw chance from speed one point eight and stealth point seven five: one point eight times one minus point two two five, divided by three, equals point four six five before countermeasure effects. The bar spans one hundred eighty six at four hundred pixels per raw chance; the second represents zero actually exercised chains in this offline diagnostic. This is an authored stochastic rule, not an empirically estimated threat probability. DOM ready first attempts Cytoscape inside a try catch that logs its missing dependency. It then initializes the D3 chart outside that catch. Missing D3 throws there and stops the later event bindings. The graph failure alone is therefore not the uncaught startup blocker. Native range input movement does not demonstrate a running simulator, functioning countermeasures or valid exports. The explanation remains a defensive abstract fixture with original failures preserved and no public deployment claim.