The Definitive Guide toAI Data Centers
Ask the GuideAboutAccount
Guide › Glossary › SL1–SL5

SL1–SL5

RAND's weights-security levels: SL1 resists amateurs, SL3 cybercrime syndicates, SL5 the most capable state operations — the guide's scale for model-theft defense.

Also written as: SL1 · SL2 · SL3 · SL4 · SL5

Where the guide uses this

11.1
Threat Model, Assets & Security Levels for AI Infrastructure
11.9
Insider Threat & Human-Layer Security
11.8
Model & Weight Protection (At-Rest, In-Transit, In-Use)
11.10
Cyber-Physical & Destructive Attacks on OT/Facility Systems

← All terms