Irregular Threat Models
Threat model matrix outlining cyber threat models from basic actors to nation-state level threats, illustrating the evolving landscape of expertise and scale upgrade from AI systems.
Model Uplift / Actor Expertise | Expertise Upgrade | Scale Upgrade | 3rd Party Tool Creation |
|---|---|---|---|
Non-Technical | Causing harm given access | Information hazard: Cyber as a potential method to cause harm | Accessibility to current SOTA capabilities |
Technical non-expert | Run end to end attack | Information hazard: Cyber as a potential method to cause harm | Automated Attack Tool |
Niche Expert (e.g. cyber crime org) | Hardened targets | Spear-phishing at scale / scaling cyber crime | Automated Attack Tool |
Cybersecurity Expert | Running an APT | Lead / target identification/ prioritization | Automated Attack Tool |
Nation-state level | Genuinely novel threat vectors | Massive scaling for Nation-state APTs | Automated Attack Tool |
AI Model (Model autonomy) | AI model self-improvement | Autonomous Replication |
Irregular Threat Models © 2024 by Irregular (Pattern Labs Tech Inc.) is licensed under CC BY-NC-ND 4.0.