Commercial maritime sensing has made it easier to build naval order-of-battle estimates from open sources. AIS provides identity and patterns but is vulnerable to spoofing and manipulation. SAR detects ships regardless of cooperation, and fusion approaches exploit mismatches between AIS and imagery to identify anomalies and “dark ships.” Commercial RF mapping can add another layer of behavioral evidence when AIS goes silent. States should counter OSINT by reducing adversary inference through emission discipline, selective disclosure, AIS governance, and better internal sharing, rather than defaulting to overclassification. Read More…
International rail networks become uniquely vulnerable at borders because critical flows concentrate into a few corridors and ports of entry, while operational interdependencies (services, rolling stock, crew) turn local constraints into network-wide delay cascades. The most effective countermeasures combine cross-border governance (shared playbooks, joint incident command, mutual aid) with technical resilience (slack capacity, modular operations, predictive monitoring, network-aware rerouting, and cyber-physical hardening), all aimed at preventing constraint overload and shortening time spent in cascade mode. Read More…
U.S. passenger rail is an open network. Airport-style checkpoints do not scale across hundreds of stations and platforms. Effective security is layered and intelligence led: visible policing and K9 presence, randomized checks, strong reporting and intel sharing, and fast incident response and recovery. The real high-leverage work sits in the cyber-physical stack that moves trains safely: signals, interlockings, dispatch, power, and communications. The post lays out a clean threat model, clarifies federal and operator roles for “critical rail,” and closes with practical guidance for travelers as well as feasible improvements for policymakers and operators. Read More…
Human intelligence is not dead in the age of ubiquitous technical surveillance (UTS), but its center of gravity is shifting. In a world where phones, cars, and cities are sensors, HUMINT has to adapt around three pressure points: • Sources are selected and developed in the shadow of pattern-of-life analytics, with elite targets either hyper-observable or deliberately off-grid. • Covers now live or die by their digital exhaust: if your pattern looks wrong to an algorithm, your legend is already burned. • Meets move from heroic “Moscow rules” streetcraft to operations that ride on, or even weaponize, the surveillance layer itself.
This post extends the Security Nexus Deep Dive episode “HUMINT Adapts to Total Telemetry” and pulls the scholarly thread tighter around UTS, cyber-enabled tradecraft, and the legal/policy environment that quietly makes all of this possible. Read More…
Export controls on GPUs and model weights absolutely shape the AI battlefield—but only where chokepoints are real, coalitions are tight, and enforcement data is exploited as aggressively as the hardware. Overreliance on broad, performance-based rules risks pushing adversaries toward harder-to-monitor paths and nudging the entire system toward fractured techno-blocs. A smarter architecture focuses narrowly upstream, leans into AI-enabled enforcement, and treats model weights as a special, high-friction case—not a magical lever. Read More…
