Supply Chain Poisoning in AI/ML Pipelines
NeuralStack | MS – AI Security Engineering Summary A modern ML system is assembled, not written. Training corpora are scraped or downloaded, weights are pulled from public hubs, and hundreds of trans
NeuralStack | MS
Latest articles
Adversarial Attacks on Multi-Modal LLMs: How Image, Audio, and Video Inputs Get Weaponized and How to Defend Against It
How to Write Engineering Content That Actually Gets Read
Four AI Security Projects You Can Ship in a Month
The Centaur Defense Model – Cooperative Defense Strategies Between Humans and AI
Identity & Trust: The Foundation of Secure AI Systems
NeuralStack | MS — AI Security Every security control eventually reduces to two questions: who is acting, and how much should they be believed. Access control answers the first question. Trust calibra
Data Provenance, Lineage & Governance as AI Security Controls
NeuralStack | MS Tech Blog – Databases & Data Engineering in AI Security Engineering, Part 4 of 4 Governance Is Not Compliance Theatre The term "data governance" carries bureaucratic connotations in
Sentinel-LLM: A Hybrid Guardrail Architecture for OWASP LLM06 – Sensitive Information Disclosure
Published on NeuralStack | MS · AI Security Engineering · June 2026 Every production LLM deployment carries a category of risk that is simultaneously well-understood in theory and chronically under-e
RAG Under Fire: Retrieval Pipeline Vulnerabilities & Indirect Prompt Injection
NeuralStack | MS Tech Blog – Databases & Data Engineering in AI Security Engineering, Part 3 of 4 The Retrieval Pipeline as a Trust Boundary Retrieval-Augmented Generation (RAG) is now the dominant a
The Hidden Attack Surface of Vector Databases
NeuralStack | MS Tech Blog – Databases & Data Engineering in AI Security Engineering, Part 2 of 4 A New Storage Primitive, A New Threat Model Vector databases emerged as a distinct infrastructure cat
Data Ingestion Attacks & AI Pipeline Security
NeuralStack | MS Tech Blog – Databases & Data Engineering in AI Security Engineering, Part 1 of 4 The Pipeline Is the Attack Surface In classical application security, the perimeter is relatively wel