Filed under ai-security.
Every note is attached to a product, a scan, or a real operational pattern we watched play out in the field. We publish when we have something specific to say — never on a content calendar.
Filtered notes
Verglos · Architecture
AI-authorship provenance, explained
Why probable AI authorship can be useful review context without becoming a vulnerability label, blame mechanism, or substitute for code evidence.
Verglos · Research
Eight patterns AI-assisted code keeps producing
Eight reviewable vulnerability shapes that recur in fast-built JavaScript and TypeScript, plus the control that should catch each one in review.
Verglos · Research
What our 300-repository TS/JS scan did and did not show
A corrected account of the Verglos scan campaign, its published artifacts, its 57.7% high-or-critical result, and the limits on interpreting scanner output.
Verglos · thesis
The security evidence layer for AI-built software
Why AI-assisted software needs reviewable security records, and which parts of Verglos' evidence model exist for builders and teams in alpha today.