Darren Death

Executive Summary Agentic workflows can materially reduce the time required to prepare and assess federal authorization packages by turning large volumes of system information into a traceable assessment record. The approach described in this paper has been implemented and tested across the full authorization workflow. The results show that a governed agentic system can perform…

Executive Summary CyberOps is a reliable, stable, fully agentic penetration-testing platform built utilizing loop engineering. It coordinates a real offensive toolkit across long-running authorized missions, maintains durable mission state, and turns tool output into structured evidence and analyst-grade findings. The platform is virtualized: the same core architecture can run on premises or in cloud infrastructure,…

For decades, the approach to building technology has operated on an implicit assumption that security could be addressed after the fact. Organizations built systems to meet functional requirements, shipped them when they worked, and addressed security through periodic assessments, patching, and monitoring. The economics of the threat environment supported this approach. Vulnerability discovery was expensive,…

Artificial intelligence is being integrated into the processes, platforms, and services that organizations depend on to deliver value. These implementations may involve systems that generate natural language in response to prompts, or systems designed to carry out defined workflows without constant human intervention. In many cases, the terminology used to describe these capabilities is applied…

Learn the key differences between generative and agentic AI, how autonomy shapes oversight and why precision in AI governance reduces operational risk. Artificial intelligence is being integrated into the processes, platforms and services that organizations depend on to deliver value. These implementations may involve systems that generate natural language in response to prompts, or systems…

AI Is Just Software, But It Is Not Just Software Artificial Intelligence (AI) is frequently portrayed as a disruptive force with the potential to revolutionize industries, optimize workflows, and enhance decision-making in ways that were often seen as unattainable. While this perspective highlights AI’s impact, it overlooks a fundamental reality. AI is still software at…

AI systems require substantial computational resources to process data efficiently. These systems generate responses that enable automation by performing tasks typically involving human intervention. They carry out complex operations, enhancing analytical capabilities and improving efficiency across various applications. Additionally, they integrate with other platforms to facilitate seamless data exchange and ensure interoperability within an organization’s…

Inaccurate outputs weaken the trustworthiness of AI systems, particularly large language models (LLMs), by generating responses that appear credible but lack accuracy. When incorrect information is presented as a fact, users may unknowingly rely on flawed outputs that seem correct but ultimately are fabricated. The underlying causes can stem from weaknesses in training data, insufficient…

Vectors and embeddings are essential components of modern AI systems, enabling the efficient processing, representation, and retrieval of complex information. These structures enhance the AI system’s ability to interpret and connect data meaningfully, leading to improved relevance and accuracy in generated responses. However, this design can also introduce vulnerabilities that compromise the reliability of the…

System prompts are essential to an AI system. Unlike user-provided prompts, these are embedded instructions that guide how a model processes input and generates output. When system prompts are exposed, they give adversaries important information that can be used to bypass access restrictions, alter decision-making rules, or extract sensitive information. Protecting these prompts is important…

As AI systems take on more complex roles, their ability to make decisions and perform tasks independently presents significant opportunities to optimize operations and foster innovation.  However, this growing autonomy also introduces new challenges, particularly when these systems function outside their intended scope.  Without careful management, such overreach can result in unintended consequences that undermine…

AI systems assist in decision-making, improve operational efficiency, and automate complex processes. However, if the output is not managed carefully, it can result in significant organizational issues, such as misleading inaccuracies and inadvertent exposure of sensitive information. These risks can undermine the reliability and effectiveness of AI systems, posing potential legal, ethical, and reputational challenges…

AI systems are heavily dependent on data, and the quality and integrity of that data significantly impact their performance. However, this reliance also creates vulnerabilities. Data and model poisoning attacks occur when the data used to train or update these systems is intentionally manipulated. Such attacks can compromise the accuracy and reliability of AI outputs,…

AI systems typically depend on various components from third-party sources, such as software libraries, pre-trained models, APIs, and hardware. These dependencies can introduce supply chain risks undermining security, reliability, and functionality. Effectively managing these risks is crucial for deploying AI systems that organizations can trust. What Are Supply Chain Risks in AI Systems? Supply chain…

Sensitive information disclosure occurs when AI systems unintentionally share private or confidential information. Organizations that utilize AI to handle sensitive data must understand this risk and take proactive steps to prevent it. This article explores how AI can inadvertently expose confidential information and outlines the measures necessary to prevent such breaches. What Is Sensitive Information…

Artificial intelligence (AI) is transforming how organizations make decisions. However, it also introduces risks that must be addressed to protect operations and data. One such risk is prompt injection, a vulnerability that can manipulate AI systems to produce harmful or unintended results. What Is Prompt Injection? Prompt injection occurs when a user submits input that…

AI security is a hot topic in today’s cybersecurity landscape due to the increasing integration of AI systems in essential areas like healthcare, finance, transportation, and national security. My company advocates for the responsible design and implementation of AI by taking an ethical, human-on-the-loop approach to AI operations. While offering immense benefits, these systems also…

As I wrote in my previous article Key Takeaways from CISA/NCSC Guidelines for Secure AI System Development, CISA’s “Guidelines for Secure AI System Development” provide a clear path for safely managing AI systems. They highlight the importance of building security into AI systems from the beginning. These guidelines help organizations deal with specific AI threats,…

What might the most damaging attacks of the future look like? The answer to the question may lie somewhere between the known patterns that attackers have established over the years, and signs that we are starting to see today. A look back It started with the sun and the moon. Solar Sunrise was discovered in…

The purpose behind this article is to pull together some interesting news from the week of 11/13/2017 as it relates to the future use of advanced drone technologies and their ability to cause serious physical harm. In the below article you are presented with a fictional story (video) where advanced drones are used to selectively…