SECSEC-001 — Model Security Vulnerabilities
Adversarial Attacks Transfer from Open-Source to Closed AI Models
4/5Sector: DefenceGeography: GlobalStage: OperateIngested: —
Executive Summary
Adversarial inputs crafted against open-weight models can bypass defences in closed-source systems, including those used in defence applications. Structured access controls offer weaker protection than assumed, exposing procured AI systems to automated manipulation.
Domain
Security & Privacy
Blindspots in model security, data poisoning, privacy leakage, infrastructure, model theft, and incident response.
Source
MIT AI Risk Repository — Risk Sources and Risk Management Measures in Support of Standards for General-Purpose AI Systems (Gipiškis2024) ↗https://airisk.mit.edu/
Could this happen in your organisation?
A Velinor AI Audit maps your active AI portfolio against the 50+ blindspots and benchmarks against documented sector failures like this one. A board-ready foresight document in 5 weeks.