Research updates, governance and privacy analysis, and public appearances across application security, AI security, and AI governance.
Research-progress updates on applying AMTD to LLM security: risk-based routing between a local and API-based model, red-team integration via Promptfoo, and a canary-string method for measuring data leakage.
A methodology update on evaluating prompt-injection-detection and guard models (via Hugging Face) as candidates for an adaptive defense pipeline.
Reflections on building a triage application for suspected AI-generated audio and video evidence -- SHA-256 integrity hashing, metadata extraction, and rule-based indicators for investigators.
A personal reflection on an AI-generated customer-service chat that wasn't disclosed as such until after the fact, proposing organizational recommendations and referencing OECD AI Principles, NIST AI RMF, ISO/IEC 42001, and the EU AI Act.
On the overlap between AI adoption, data privacy obligations, and security controls.
On governed AI adoption and what organizations need in place before deploying AI systems.
Reflections from a security operations center visit: SIEM tuning, vulnerability management workflows, security signals, and governed AI adoption in practice.
Interviewed by Orji H. Chika on cybersecurity education and how students and early-career professionals are positioning themselves. One segment covers Moving Target Defense as an LLM security mechanism.
The AMTD research and Digital Forensics work above are covered in full depth as their own case studies:
AMTD case study → · Deepfake Fraud Evidence Triage case study →