SECSEC-001 — Model Security Vulnerabilities
Large Language Models Providing Harmful Scientific Instructions
4/5Sector: OtherGeography: GlobalStage: OperateIngested: —
Executive Summary
LLMs demonstrated capability to generate step-by-step instructions for conducting dangerous scientific experiments, constituting a direct dual-use risk. Organisations deploying or procuring such models face regulatory exposure and reputational liability without robust capability evaluation and content governance frameworks.
Domain
Security & Privacy
Blindspots in model security, data poisoning, privacy leakage, infrastructure, model theft, and incident response.
Source
MIT AI Risk Repository — Cataloguing LLM Evaluations (InfoComm2023) ↗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.