Insights

Thoughts & Writing

Practical perspectives on security architecture, emerging threats, and building resilient programs.

Penetration Testing

OT and ICS Penetration Testing Is Not a Network Assessment

Industrial control systems run refineries, water treatment plants, and power substations. Testing their security requires a methodology that looks nothing like a standard network assessment, and applying the wrong one can take down a production line.

Jul 22, 20265 min read
Read
Compliance & GRC

The Vendor Security Questionnaire Is Not a Risk Assessment

Most companies send security questionnaires to vendors as a form of due diligence. What they get back is a compliance artifact, not an assessment of risk. Here is what the questionnaire is actually measuring and what to do instead.

Jul 20, 20265 min read
Read
Security Architecture

Detection Engineering Is Not a SIEM Subscription

Most organizations treat a SIEM license as a detection program. It is not. Detection engineering is the discipline of writing, testing, and maintaining detection logic tuned to your environment, and vendor-supplied rules are not a substitute for it.

Jul 18, 20265 min read
Read
AI/ML Security

Embeddings Are Not Anonymization

Turning your company's sensitive data into vector embeddings does not make it safe. Embeddings are more reversible than most practitioners assume, and vector stores are typically missing the access controls your relational databases take for granted.

Jul 16, 20265 min read
Read
Penetration Testing

Your Web Application Penetration Test Is Probably Scoped Wrong

Most web application penetration tests are commissioned with the right intention and structured in a way that makes them significantly less useful than they could be. Scoping is where value gets lost, before the tester ever opens a browser.

Jul 14, 20265 min read
Read
Incident Response

Threat Hunting Is Not the Same Thing as Waiting for Alerts

Security operations teams that rely entirely on alerts are making a bet that their detection coverage is complete. It is not. Threat hunting is what you do when you assume the attacker is already there and set out to prove it, before the alert fires.

Jul 12, 20265 min read
Read
Compliance & GRC

PCI DSS 4.0 Is the Compliance Update That Actually Changed Something

PCI DSS 4.0 became mandatory in March 2024 and it is not a cosmetic update. MFA requirements expanded significantly, targeted risk analyses became required work, and organizations that treated the transition as a documentation exercise are finding out what they missed during their first formal assessment.

Jul 10, 20265 min read
Read
Security Architecture

Kubernetes Security: The Defaults Will Get You

Kubernetes ships with defaults that prioritize getting things running over keeping them secure. Most clusters in production are running on those defaults. Here is what that actually looks like from an attacker's perspective and what a reasonable baseline requires.

Jul 8, 20265 min read
Read
Incident Response

Your Incident Response Playbook Assumes There's a Disk to Image

Cloud environments violate most of what traditional incident response assumes. There is no disk to image, no cable to pull, and the evidence that exists is ephemeral. Here is what IR actually looks like when the incident is in AWS.

Jul 6, 20265 min read
Read
AI/ML Security

Your Internal AI Assistant Has Your Data Memorized. And It Can Recite It.

Organizations are rushing to fine-tune AI assistants on internal documents, support tickets, and HR data. The model does not just learn from that data - it memorizes parts of it. Training data extraction is a documented attack technique, and most internal AI deployments have not accounted for it.

Jul 4, 20265 min read
Read
Compliance & GRC

When You Buy a Company, You Buy Its Security Debt

M&A due diligence covers financials, legal, and operations with rigor. Security usually gets a questionnaire. The acquiring company finds out what was actually on the target's infrastructure after the deal closes, which is the worst possible time to find out.

Jul 2, 20265 min read
Read
Penetration Testing

Purple Teaming: What Happens When Red and Blue Finally Work Together

Purple teaming is what happens when you stop treating offensive and defensive security as separate departments running separate exercises. It is the fastest way to find out whether your detection coverage actually works before a real attacker demonstrates that it does not.

Jun 30, 20265 min read
Read
Security Architecture

Certificate Management: The Invisible Outage Waiting to Happen

Expired TLS certificates cause production outages that have nothing to do with attackers. Most organizations manage thousands of certificates with spreadsheets and panic. Here is what certificate management looks like when it works and why it almost never does.

Jun 28, 20265 min read
Read
Incident Response

Business Email Compromise Response Has a Different Clock

Business email compromise does not look like a security incident until the wire is already gone. The response runs on a different clock than every other incident type, and the urgent action in the first two hours is financial, not technical.

Jun 26, 20265 min read
Read
AI/ML Security

Agentic AI Has Tool Access Now. That Changes Everything.

Most AI security thinking was built around chatbots that generate text. Agentic AI systems can query databases, send emails, execute code, and spawn sub-agents. The attack surface is different, the failure modes are different, and the security frameworks most teams are applying are not adequate.

Jun 24, 20265 min read
Read
Penetration Testing

Cloud Penetration Testing Requires a Different Playbook

Most organizations buy cloud penetration tests and get network assessments with cloud-flavored findings. The attack surface in AWS, Azure, and GCP is API-driven and IAM-centric, and the methodology that actually finds the dangerous paths looks nothing like a port scan.

Jun 22, 20265 min read
Read
Compliance & GRC

Cyber Insurance Is Not a Backup Plan

Most organizations treat cyber insurance as a financial backstop for incidents their security program failed to prevent. Insurers increasingly see it differently. The gap between those two perspectives is where claims get denied.

Jun 20, 20265 min read
Read
Incident Response

When Ransomware Lands: The Decisions That Matter in the First 48 Hours

Ransomware response is not one decision. It is a sequence of decisions that interact, each made under time pressure with incomplete information. Most organizations have not thought through any of them before needing to make all of them at once.

Jun 18, 20265 min read
Read
Security Architecture

The Build Pipeline Is the New Perimeter Nobody Is Defending

Modern attackers bypass hardened production environments by targeting what generates them - the build pipeline, package dependencies, and CI/CD tooling. Most engineering organizations are running completely blind to this attack surface.

Jun 17, 20265 min read
Read
Penetration Testing

How Internal Penetration Tests End at Domain Admin

The most common path to domain administrator in an internal network assessment is not a zero-day exploit. It is a series of configuration gaps that exist in almost every Active Directory environment. Here is how that path works and what it looks like from the other side.

Jun 16, 20265 min read
Read
AI/ML Security

How Attackers Beat Your Machine Learning Security Controls

Machine learning has become the backbone of modern security detection. Attackers have been quietly learning how to beat it. Here is what adversarial ML looks like in practice and why layered detection still matters.

Jun 15, 20265 min read
Read
Security Architecture

The API Endpoints Nobody Is Watching

Most organizations have no accurate inventory of their APIs, no consistent authentication on internal endpoints, and at least one deprecated version still responding to requests. The API layer is where modern attacks increasingly land, and it remains the least consistently governed part of the stack.

Jun 14, 20265 min read
Read
Compliance & GRC

The SOC 2 Report Nobody Actually Reads

Collecting SOC 2 reports from vendors is standard practice. Actually reading them is rare enough to be exceptional. The gap between the report and the risk decision it is supposed to inform is where most vendor security programs quietly fail.

Jun 13, 20265 min read
Read
Incident Response

Are They Really Gone? The Eradication Question Nobody Answers Well

Declaring an incident closed after patching the entry point is not the same as knowing the attacker is out. Most organizations conflate eradication with containment and call it done before they have answered the hardest question in incident response.

Jun 11, 20265 min read
Read
Penetration Testing

Red Team Versus Penetration Test: You're Probably Buying the Wrong One

Every year, organizations spend serious money on red team engagements when what they actually need is a penetration test. The terms get used interchangeably in vendor proposals, which serves the vendor well and the buyer poorly. Here's what each one actually measures.

Jun 11, 20265 min read
Read
Security Architecture

Your Secrets Are Already in Version Control

A developer commits a credential, notices it three days later, rotates the key, and closes the ticket. What nobody checks is whether anyone used it in between. Here is where secrets actually live and how to stop losing them.

Jun 10, 20265 min read
Read
Incident Response

The Post-Incident Review Nobody Follows Through On

Most organizations run post-incident reviews. Very few of them produce meaningful change. The problem is not that teams skip the meeting. It is that they mistake documenting what happened for learning from it.

Jun 9, 20265 min read
Read
Incident Response

Who Else Needs to Know Right Now

In the first hour of an incident, your team is focused on what is happening technically. Nobody is thinking about legal counsel, the regulatory clock, or the customer whose data might be in the pile. That notification problem is its own fire, and it starts burning the moment you decide to handle it later.

Jun 9, 20265 min read
Read
Incident Response

Your Incident Response Plan Is a PDF Nobody Has Read

Most organizations have an incident response plan. Most of those plans live in a document that was last opened during the audit that required it. The gap between having a plan and being able to execute one is where incidents go to become disasters.

Jun 9, 20265 min read
Read
Incident Response

Containment Is a Decision, Not a Button

The first hour of an incident is full of pressure to do something, and the most tempting something is to pull the plug. But containment trades away evidence, visibility, and uptime, and the responders who do it well treat it as a decision with a cost, not a reflex.

Jun 8, 20265 min read
Read
AI/ML Security

Shadow AI: Your Team Already Pasted the Roadmap Into a Chatbot

While your governance committee debates an AI usage policy, your employees have quietly adopted a dozen AI tools and fed them everything from source code to customer lists. Shadow AI is the new shadow IT, and it moves faster.

Jun 7, 20265 min read
Read
Compliance & GRC

The Security Exception That Was Supposed to Be Temporary

Every organization has a stack of approved security exceptions that were meant to last 30 days. Some of them are five years old. Here is how exception debt accumulates, why nobody notices, and what it costs when someone finally does.

Jun 5, 20265 min read
Read
AI/ML Security

Your AI Model Has No Idea What Year It Is

Large language models are trained on data with a cutoff date and then deployed for months or years afterward. The gap between what the model knows and what is currently true is a security problem that most AI deployments have not accounted for.

Jun 4, 20265 min read
Read
AI/ML Security

Deepfake Attacks Are Here and Your Security Team Isn't Ready

In 2024, a finance employee in Hong Kong transferred 25 million dollars to criminals after a video call with what appeared to be his CFO. Everyone on the call was fake. Here's what AI-generated social engineering looks like now and what you can actually do about it.

Jun 3, 20265 min read
Read
Penetration Testing

The Low-Hanging Fruit That's Always There

After hundreds of penetration tests across dozens of industries, a pattern emerges. Different organizations, different sizes, different maturity levels. Same findings, every time. Here's the short list of things that are almost always wrong and why they stay that way.

Jun 2, 20265 min read
Read
Compliance & GRC

MFA Is Not the Finish Line

Enabling multi-factor authentication is one of the most effective controls you can deploy. It is also not nearly enough on its own, and attackers figured that out a while ago. Here's what MFA actually stops, what it doesn't, and what comes after.

Jun 1, 20265 min read
Read
Security Architecture

You Have 40,000 Vulnerabilities. Here's How to Care About the Right Ones.

The average enterprise vulnerability scanner returns tens of thousands of findings. Treating them all equally is operationally impossible and strategically wrong. Here's how to build a prioritization model that focuses your team on the vulnerabilities that are actually going to get you.

May 31, 20266 min read
Read
Penetration Testing

The Phishing Email That Got Past Me

I write phishing emails for a living. I've sent thousands of them to employees as part of authorized engagements. I know every trick in the playbook. And one got me anyway. Here's what happened and why it matters.

May 29, 20265 min read
Read
Security Architecture

Your Cloud Probably Has a Public S3 Bucket. You Just Don't Know About It Yet.

The most common finding in cloud security assessments isn't an exotic zero-day. It's a checkbox that defaulted to public, an IAM role that was supposed to be temporary, and a storage bucket named after an internal project sitting fully exposed to the internet.

May 28, 20266 min read
Read
Compliance & GRC

Security Awareness Training Doesn't Work. Here's What Does.

Every year, employees click through the compliance module. Every year, people still get phished. The annual training checkbox is not a security control. Here's what actually changes human behavior in ways that hold up under real attack conditions.

May 27, 20265 min read
Read
Compliance & GRC

Your Password Policy Is a Work of Fiction

Eight characters, one uppercase, one number, one symbol. Changed every 90 days. Completely useless. Here's why your password policy is training users to create worse passwords, and what to do instead.

May 25, 20266 min read
Read
Penetration Testing

I Walked Into Your Building With a Box of Donuts

Physical penetration testing is the part of security assessments nobody wants to think about because the findings are deeply embarrassing. A look at how alarmingly easy it is to walk past $2 million worth of security technology with a smile and a pastry.

May 23, 20265 min read
Read
Security Architecture

Zero Trust Architecture: Beyond the Buzzword

Zero Trust has become one of the most overused terms in enterprise security. Here's what it actually means and what it takes to implement it in a way that reduces real risk.

May 20, 20265 min read
Read
Security Architecture

Zero Trust in Practice: Building a Roadmap That Sticks

Most Zero Trust programs stall after the first phase. The reason is almost never technical. Here's how to build a roadmap that survives contact with your organization.

May 15, 20265 min read
Read
AI/ML Security

AI-Powered Threat Detection: Separating Signal from Noise

Machine learning in security operations creates as many false positives as it prevents. Here's how to tune your detection models and build analyst workflows that scale.

May 10, 20265 min read
Read
AI/ML Security

LLM Security: The Attack Surface No One's Ready For

Large language models introduce a new class of vulnerabilities that traditional security controls weren't designed to catch. Here's what security teams need to understand now.

May 5, 20265 min read
Read
Compliance & GRC

NIST CSF 2.0: What Changed and What It Means for Your Program

The updated framework introduces a new Govern function and expands scope beyond critical infrastructure. A practitioner's guide to mapping your existing controls to the new structure.

April 28, 20265 min read
Read
Penetration Testing

What Really Happens During a Penetration Test

A penetration test is not a vulnerability scan with a human attached. Here's what a real engagement looks like from scoping to final report and what separates useful findings from checkbox compliance.

April 20, 20265 min read
Read