CHEEWAN.AI // DEFENSIVE FRAMEWORK v3.0 // EMERGING AI THREATS
A 6-Domain operating model for emerging AI security threats
17 named threats catalogued across both directions of the AI security threat model. The framework morphs by lens; threats can be selected to focus the view on a specific incident.
SOURCE-VALIDATED
About →PRIMARY LENS · CHANGES LAYOUT
Each lens morphs the diagram layout · Dimensions filter the threat population without changing the shape
17 of 17 threats match
DIRECTION
ATTACKER SOPHISTICATION
AI CAPABILITY
EVIDENCE QUALITY
SYSTEM VIEW
LIVE
Threat hub
Defensive domains
SELECT A THREAT BELOW ↓
DOMAIN D02
Threat-Model
Use named public catalogues so the SOC and audit speak one language. Inventing taxonomies dilutes the signal.
OWNER
Threat Intel Lead
FRAMEWORKS
3 mapped
ADOPTION
Days 31-60
CONTROLS · 5 · ALL CITED
- 01AML.T0061 AI Agent Tools — abuse of agent toolset
- 02AML.T0062 Exfiltration via AI Agent Tool Invocation
- 03LLM01 Prompt Injection — primary GTG-1002 vector
- 04LLM06 Excessive Agency — most critical for agents
- 05ASI10 Rogue Agents — behavioural detection required
THREAT CATALOG · 17 SHOWN↓ Click any card to focus the diagram on a specific threat
PHANTOM TEAL · DEFENSIVE FRAMEWORK v3 · CHEEWAN.AI
● NIST ·● MITRE ·● OWASP ·● CISA