Can We Investigate AI as Fast as We Train It?
A digital forensic perspective on what it means to investigate model behaviour, test causal explanations, and limit deployment authority when the evidence remains incomplete.
Articles and analysis
Evidence, technology, and the conclusions we ask them to support.
Independent writing on digital forensics, AI assurance, governance, and professional judgment.
Latest article
When Malware Knows You're Investigating It
How AI-assisted concealment could change memory acquisition, and what investigators need to distinguish missing artifacts from supported findings.
Read the latest articleStop the Ridiculousness
Why unauthorized AI intrusions should produce evidence, accountability, and correction rather than exaggerated claims about machine capability.
When AI Safety Becomes a Barrier to Entry
How frontier AI safety rules can become an incumbent-controlled gate that open development has no defined path to cross.
The 10% AI Extinction Claim Has Not Earned Its Authority
Why an unexplained forecast of a greater-than-10% chance of AI killing every human has not earned the evidentiary authority attached to it.
Stop Calling Everything Digital Forensics.
Why tools, job titles, and the DFIR label do not make an investigation forensic without an evidentiary process built to withstand independent challenge.
No One Directed It. Someone Is Still Accountable.
Why delegated autonomy cannot become an escape hatch from human accountability when AI systems cause harm.
A Watermark Is Not a Finding
Why an AI watermark is evidence of technological provenance, not proof of authorship, misconduct, or a policy violation.
The Wrong Question About AI and Authorship
Why AI involvement is too crude a test for authorship, and why meaningful human creative agency is the better question.
What Logic Gates Taught Me About Artificial Intelligence
How building logic circuits shaped an approach to artificial intelligence, assurance, and justified trust.
What You Can Defend
How generative AI is weakening traditional signals of expertise, and why defensibility now matters more than polished production.
Smarter Than All of Us
What happens when a consequential claim about AI enters planning and governance before anyone has defined what would establish it?
AI Agents Are Not the Problem. Our Security Posture Is.
Documented agentic AI failure modes, approval-gate weaknesses, supply-chain compromise, and forensic readiness.