Insights
White papers, technical analysis, and commentary from the KAIperShield team — on architecture-level cyber resilience, trustworthy secure systems, and the policy landscape shaping mission-critical security.
Research & Technical Analysis
In-depth technical papers from our team grounded in NIST SP 800-160 and decades of systems security engineering experience.
A Multidimensional Protection Strategy for Building Mission-Resilient Systems in the Age of Frontier AI
Frontier AI models capable of autonomous, multi-step reasoning and code generation create two distinct problems for mission-resilient system design: AI as a threat-actor capability multiplier, and AI as a system component whose statistical behavior resists the specification and verification classical trust models require. This paper argues that NIST SP 800-160 supplies the correct engineering foundation — but that the foundation alone is not sufficient — and proposes a five-layer strategy spanning systems engineering, AI-specific threat modeling, operational tempo, data and model integrity, and governance. Sufficiency against frontier AI is structural, not behavioral: an architecture must hold regardless of what a model intends, chooses, or is instructed to do.
Engineering Mission Resilience in the Age of Frontier AI
Mission-critical systems fail at the seams — between system components, teams, and authorities. A siloed cybersecurity approach can improve individual defenses while leaving the relationships among components, authorities, services, and physical processes insufficiently constrained, and frontier AI widens those seams. Faster detection does not fix excessive authority, a bypassable boundary, or a shared dependency that defeats recovery. Drawing on NIST SP 800-160 and the Engineering for Compromise methodology, this paper presents a loss-driven approach built on five commitments: start from mission losses rather than the tool inventory; assume selected elements are compromised and ask what each could cause; constrain consequences outside the compromised element; engineer recovery before the incident; and keep the evidence for every protection claim.
Trustworthy Secure Operational Technology Systems
The draft fourth revision of NIST SP 800-82 provides an essential and comprehensive foundation for managing cybersecurity risk in operational technology. Yet security controls, practices, and reference architectures alone cannot establish that an OT system will remain trustworthy secure when components fail, credentials are stolen, maintenance pathways are abused, supply chains are compromised, or adversaries obtain a foothold. Trustworthy secure systems concepts from NIST SP 800-160 add the missing engineering dimension: define unacceptable loss, apply the Security Design Order of Precedence, engineer for compromise, and produce evidence that justifies confidence in system-level claims. The result is a constructive extension from protecting OT assets to engineering OT systems whose architecture bounds loss and preserves essential services.
Industrial-Scale AI Distillation
A September 2026 joint NSA, CISA, and FBI advisory describes industrial-scale campaigns by China-based AI companies to extract proprietary capabilities from U.S. frontier models — millions of requests across distributed accounts, cloud providers, third-party aggregators, and proxy transfer stations that transform ordinary inference access into a strategic collection mechanism. This article argues that malicious distillation is not merely account abuse, fraud, or excessive API consumption, but an unauthorized transfer of computational capability assembled from individually plausible interactions. Effective defense requires more than stronger authentication and static rate limits: define unacceptable losses, establish security domains, mediate every consequential interaction, limit persistence and sharing, detect coordinated anomalies across organizational boundaries, and produce evidence that the architecture constrains extraction under realistic adversary behavior.
The Art of Systems Security
Sun Tzu's The Art of War and NIST SP 800-160 arise from radically different eras, purposes, and disciplines. Yet both place extraordinary value on preparation, knowledge, positioning, disciplined control, adaptation, and the deliberate shaping of conditions before adversity is encountered. This paper uses Sun Tzu's strategic ideas as an interpretive lens for the thirty principles for trustworthy secure design — claiming no historical lineage or one-to-one mapping, but showing how ancient strategic insights make modern engineering ideas more accessible. The central argument: trustworthy secure systems are not produced by accumulating controls after design decisions have been made, but by arranging trust, authority, information flow, interfaces, dependencies, failure behavior, and recovery so that unacceptable loss is made difficult to reach. Sun Tzu sought advantage before battle; systems security engineering seeks protection before compromise.
Assurance and Trustworthy Secure Systems: Why Evidence Is the Basis for Trust
Trustworthy secure systems cannot be established by assertion, reputation, compliance, or the presence of security functionality alone. NIST SP 800-160 makes assurance the evidentiary foundation for trustworthiness: justified confidence that security claims have been achieved, derived from objective evidence that reduces uncertainty to an acceptable level. This paper develops assurance as the unifying concept across trustworthy secure design, adequate security, engineering for compromise, component trustworthiness, and resilience. It distinguishes trust from trustworthiness, explains why the required level of trust must be commensurate with consequence, and gives particular attention to opaque or black-box components, where limited visibility constrains the evidence obtainable and therefore the trustworthiness that can responsibly be claimed. The central proposition is simple: trust should never exceed the assurance that the available evidence can justify.
Trustworthy Secure Design: Engineering Security from the Inside Out
NIST SP 800-160, Volume 1 offers a timely engineering response to frontier AI, complex supply chains, autonomous agents, and accumulated technical debt: treat security as an emergent property of a system; begin with intended behaviors, system states, and the conditions that can lead to loss; and apply the Security Design Order of Precedence (SecDOP) to eliminate or reduce loss potential through design before relying on engineered features, monitoring, procedures, or training. The central argument is simple — organizations cannot operationally compensate forever for architectures that were never designed to control asset loss. The most consequential security decisions are often made before a security control is selected.
Engineering the Cyber Defense Surge
OpenAI's open letter, A Call for Collective Action on Cyber Defense, warns that AI-enabled attacks will become more widespread and sophisticated and that status quo security is inadequate. Its recommendations — least privilege, defense in depth, continuous testing, containment, recovery, verified fixes — are directionally sound. This article makes a complementary argument: much of the engineering foundation needed to answer that call already exists in NIST SP 800-160. The central problem is not the absence of a conceptual foundation but the persistent implementation gap between what the community has known how to engineer and what organizations have actually funded, acquired, built, and sustained. AI changes the urgency, speed, scale, and economics of cyber conflict — it does not invalidate the principles.
Adequate Security: An Engineering Answer to an Impossible Question
Security discussions often begin with threats, controls, products, frameworks, or compliance requirements. The more fundamental engineering question is simpler: how much security does a system actually need? NIST SP 800-160, Volume 1 answers through the concept of an adequately secure system — neither perfect security nor the maximum technically possible, but an evidence-based judgment that the system achieves at least a minimum tolerable level of security and is as secure as reasonably practicable when balanced against cost, schedule, technical-performance, operational, and mission constraints. The central proposition: the right amount of security is the amount that can be justified by stakeholder consequences, system context, engineering trade space, and evidence — not by intuition, compliance alone, or the pursuit of an unattainable absolute.
AI Data Centers: The New Center of Gravity
As frontier AI increasingly supports intelligence, command and control, cyber operations, logistics, autonomous systems, and critical infrastructure, the facilities that provide that computational foundation acquire strategic significance of their own. This article argues that in a major conflict with a technologically sophisticated adversary, concentrations of advanced AI compute are likely to become a new strategic center of gravity. Drawing on U.S. military center-of-gravity doctrine, Sun Tzu, Engineering for Compromise, RAND's Secure Inference Data Center concept, and NIST SP 800-160, it proposes treating critical AI infrastructure as an integrated national-security system — computation, facilities, power, cooling, communications, personnel, supply chains, and recovery — with the security design principles applied recursively from the cyber architecture into the facility. The objective is not an invulnerable data center, but a national AI capability engineered so that compromise of individual facilities does not automatically become strategic mission loss.
Engineering for Compromise: Designing Systems That Remain Trustworthy Secure Under Adversity
A systems security engineering strategy that assumes selected system elements may fail, become malicious, or fall under adversary control — and asks whether the system can continue to prevent unacceptable loss. Introduces compromise containment boundaries, propagation analysis, and compromise margin, grounded in NIST SP 800-160.
Engineering for Compromise: Briefing Deck
A briefing-format companion to the white paper above, walking through the engineering question at the center of EfC — if this element were fully controlled by an adversary, what authority and reach would it inherit? — the causal chain from component compromise to unacceptable loss, compromise containment boundaries and the design principles that produce them, compromise margin and how to test whether barriers are truly independent, and a ten-step methodology applied across the life cycle. The closing argument: a trustworthy secure system is not one in which nothing can be compromised, but one engineered so that compromise does not automatically become catastrophe.
Engineering Trustworthy Secure AI Infrastructure
RAND's Secure Inference Data Centers report proposes a purpose-built architecture for protecting strategically important AI models and inference data against an exceptionally capable nation-state adversary. Its strongest contribution is not a collection of controls but a vertically integrated security argument. This paper argues that NIST SP 800-160 provides the design foundation for that argument and the engineering discipline to sustain it across the entire system life cycle — domain separation, minimality, mediated access, protective defaults, protective failure, detection, recovery, and evidence operating as a causal chain in which unauthorized behavior is progressively eliminated, constrained, detected, contained, and recovered from.
Engineering Cryptographic Resilience
Cryptographic strength — measured by mathematical hardness and key length — can no longer be assumed durable: AI-accelerated cryptanalysis has produced structural degradation in candidate post-quantum algorithms and accelerated attacks against symmetric schemes. This paper presents hyper crypto agility: a policy-governed, standing operational capability to discover, assess, deprecate, replace, and verify classical, post-quantum, and multi-jurisdictional cryptographic primitives faster than adversaries can operationalize exploits. Synthesizing NIST SP 800-160 design principles with intent-based cryptographic abstraction, it makes the case for moving from episodic migration projects to continuous, automated lifecycle management — turning cryptographic change from a disruptive failure event into a controlled, verifiable system property.
Trustworthy Systems Need Trustworthy Parts
NIST SP 800-160 holds that system-level trustworthiness cannot be assumed — it must be established from evidence about the composed elements before any composition argument begins. This paper argues that SP 800-160 depends on substantiated component trustworthiness, and that the Common Criteria (ISO/IEC 15408) is the principal standardized mechanism SP 800-160 itself cites for producing it. Maps six design principles to Common Criteria mechanisms, shows that domain separation — the architectural core of defense against autonomous AI-driven attacks — is only as trustworthy as the components enforcing it, and treats frontier AI as a component whose non-deterministic behavior resists that evaluation model.
Defending Against Mythos-Class Attacks
The next generation of AI-driven cyber threats — autonomous zero-day exploitation, multi-stage attack chains, rapid low-cost execution — represents a fundamental shift to machine-speed cyber warfare. Traditional controls-based security cannot stop adversaries whose tools are explicitly designed to bypass controls. This paper makes the case that structural design principles from NIST SP 800-160 are the essential foundation: domain separation and seven complementary principles creating an interlocking architectural defense that imposes cost at every phase of the cyberattack lifecycle.
PQC Transition: A Systems Engineering Imperative
PQC migration is framed as a cryptography problem — but that framing is insufficient. Replacing algorithms without addressing underlying system architecture installs gold-standard cryptographic protection on a house of cards. Mission resilience requires NIST SP 800-160 systems engineering discipline, not algorithm migration alone.
PQC Transition: A Systems Engineering Imperative
A briefing-format companion to the white paper above. Walks through why algorithm compliance is not the same as system trustworthiness, the eight NIST SP 800-160 security design principles that govern a defensible PQC architecture, and how to use the migration as a forcing function to fix decades of technical debt.
Protecting Mission Critical Systems: The Need for a Shift in Culture, Strategy, and Process
Published in INCOSE's INSIGHT magazine. Proposes a trustworthy secure systems engineering approach — grounded in NIST SP 800-160 — as an alternative to compliance-based RMF for space systems, and presents early results from the NASA/JPL SunRISE pilot, where the SP 800-160-engineered digital twin detected a data-tampering attack in minutes versus weeks under the traditional approach.
Space System Survivability Against AI-Driven Cyberattacks
Argues that AI-enabled "Mythos-class" attack tools — capable of autonomous vulnerability discovery and exploit generation at machine speed — make reactive patching insufficient for space systems that can't be rapidly reconfigured once on orbit. Shows how domain separation and eight complementary NIST SP 800-160 design principles constrain adversarial movement across the ground, link, and space segments, and presents empirical results from the NASA SunRISE pilot.
Space System Survivability: Briefing Deck
A briefing-format companion to the white paper above, walking through the changed economics of AI-driven exploitation, the three-segment space system attack surface, domain separation as the architectural foundation for defense, and the SunRISE pilot results — including detection times improving from weeks to seconds.
Engineering with the Security Design Principles
A series of articles, each taking a deep dive into one of the NIST SP 800-160 security design principles — examining it as a systems security engineering discipline rather than a control or a networking technique. Additional articles are coming soon.
Domain Separation
Often interpreted as network segmentation, domain separation is fundamentally about controlling influence — determining what a compromised system element can reach, affect, command, disclose, or cause elsewhere in the system. Article 1 examines the principle through the lens of systems security engineering and Engineering for Compromise, asking where an element's influence stops once it becomes compromised. The central proposition: a meaningful security domain is not defined merely by where components are placed, but by a defensible limit on influence — and a boundary is trustworthy only to the extent that there is evidence that compromise in one domain cannot produce prohibited effects in another.
Mediated Access
NIST SP 800-160 states the principle directly: all access to and operations on system elements are mediated. The statement is compact, but its implications extend far beyond user logon and conventional access control. Mediation determines whether every consequential interaction — human-to-machine, machine-to-machine, cyber-to-physical, administrative, maintenance, recovery, and supply-chain initiated — is subjected to an explicit policy decision and dependable enforcement before it can affect a protected asset or system state. Article 2 examines the principle through the lens of systems security engineering and Engineering for Compromise. The central proposition: a system cannot credibly claim control over its behavior when consequential interactions can occur outside its mediation architecture. A trustworthy mediator does more than decide who may enter — it constrains what may be done, to what, under which conditions, for how long, and with what evidence.
Least Privilege
NIST SP 800-160 states that each system element is allocated the privileges necessary to accomplish its specified functions but no more — a statement that applies to hardware, firmware, software, and personnel, and whose implications extend well beyond restricting administrator accounts. Privilege determines what a system element can reach, change, invoke, delegate, disable, and cause; every unnecessary privilege expands the consequences of error, misuse, failure, or compromise. Article 3 examines the principle through the lens of systems security engineering and Engineering for Compromise. The central proposition: authority must be justified by a specified function and bounded by the consequences of its exercise — making least privilege both an allocation discipline and a constraint on the capabilities an adversary inherits when a system element becomes untrustworthy.
Least Functionality
NIST SP 800-160 states that each system element has the capability to accomplish its required functions but no more. The implications extend beyond disabling unused services: every unnecessary function introduces additional complexity — behavior, interfaces, dependencies, or state — that can increase susceptibility, vulnerability, and the potential for unacceptable asset loss. Article 4 examines the principle through the lens of systems security engineering and Engineering for Compromise. The central proposition: functionality must be justified by the intended purpose of the system element and constrained in the realized system. Where unnecessary functions cannot be removed, they must be disabled, disarmed, placed in a safe mode, or subject to dependable mediation — sufficient mission capability with no unjustified functional exposure.
Commentary & Analysis
Short-form perspectives from the KAIperShield team on the policy and technical landscape. Follow RONROSSECURE on LinkedIn for the latest.
Trustworthy Systems Need Trustworthy Parts
We cannot build trustworthy systems from components whose trustworthiness we merely assume. Trust must be substantiated with evidence — not simply assumed. This article examines the relationship between NIST SP 800-160 and the Common Criteria (ISO/IEC 15408) and argues a precise distinction: SP 800-160 does not require the Common Criteria as the only possible path, but it does require substantiated component trustworthiness. Domain separation is only as trustworthy as the components enforcing the boundaries.
Read on LinkedIn ↗Mission Resilience in the Age of Frontier AI
Frontier AI doesn't just attack systems faster — it becomes a system component whose behavior no one can fully specify. Good intentions and successful evaluations aren't enough to trust it. Sufficiency has to be structural, not behavioral. It's essential for mission resilience, and for national and economic security. We need to act now.
Read on LinkedIn ↗An Executive Order for Trustworthy Systems?
Are we paying as much attention to developing trustworthy secure systems as we are to addressing cryptography in the quantum computing age? This article discusses why both are important to long-term national and economic security.
Read on LinkedIn ↗Defending Against Mythos-Class Attacks
Next generation, AI-informed cyber weapons are here. It's time to change the primary focus from vulnerability detection (penetrate and patch) to building trustworthy secure systems that are mission resilient.
Read on LinkedIn ↗PQC Transition: A Systems Engineering Imperative
PQC migration alone will not protect mission-critical systems and high-value assets. Unless and until we have a full court press on reengineering current system architectures to ensure they are trustworthy secure and mission resilient, adversaries with next generation AI attack tools will continue to have the capability to inflict severe or catastrophic damage on the U.S. critical infrastructure.
Read on LinkedIn ↗