Knowledge Base Regulation Tracker Standards & Frameworks OSS Directory AI Incident Database Research Paper Library Case Studies Society Articles News Jobs About Charter Sign inBecome a Member
Home/Research Paper Library/Securing AI Model Weights

Securing AI Model Weights

RAND's framework of five security levels for model weights, adopted by frontier developers' safety frameworks and government guidance as the reference for weight protection.

weight securityrandsecurity levels

Research question

What would it take to protect frontier model weights from capable attackers?

Key findings

  • 38 attack vectors across nine categories, from insider threats to supply chain - Five security levels (SL1-SL5) matched to attacker capability categories up to top state actors - No current lab reaches SL5; achieving it needs years of investment

Sources

PreviousGet my drift? Catching LLM Task Drift with Activation DeltasNextSecuring AI Model Weights: Preventing Theft and Misuse of Frontier Models

Contribute

Spot an error, or know a record we should add? Members can propose additions and submit corrections.

Submit researchReport a correction or request removal

Related in Research Paper Library

Forgeable Confirmation in Automated Computer Security Testing: Deterministic Rules versus AI Judges

AI is increasingly used to automate computer security testing, and the tools must decide for themselves whether an attack succeeded. A finding that a deterministic rule confirms by observation is reported as fact,…

Model weight security2026

When Label Noise Meets Class Imbalance: A Robust Framework for Android Malware Family Classification

Machine learning methods for Android malware family classification have achieved high accuracy, but their application is hindered by two major challenges. First, the widely used code obfuscation severely disrupts the…

Model weight security2026

An LLM-Assisted AutoML Framework for Intrusion Detection in IoT Networks

Internet of Things (IoT) systems are increasingly deployed in smart homes, transportation, energy systems, and critical infrastructure. This broad connectivity improves service intelligence, but also enlarges the…

Model weight security2026