rule AATMF_T1_InstructionOverride {
meta:
tactic = "T1"
technique = "T1-AT-001"
severity = "HIGH"
description = "Detects instruction override injection patterns"
strings:
$s1 = /ignore\s+(previous|above|all|prior)\s+(instructions?|rules?|prompts?)/i
$s2 = /you\s+are\s+now\s+(DAN|evil|unrestricted|jailbroken|unfiltered)/i
$s3 = /\[(SYSTEM|INST|SYS)\]/i
$s4 = /<\|?(system|im_start|im_end|endoftext)\|?>/i
$s5 = /BEGIN\s+(OVERRIDE|NEW.INSTRUCTIONS|JAILBREAK)/i
$s6 = /(admin|root|developer)\s*(mode|access|override)/i
condition:
any of them
}rule AATMF_T2_EncodingEvasion {
meta:
tactic = "T2"
technique = "T2-AT-001 through T2-AT-005"
severity = "MEDIUM"
strings:
$base64 = /[A-Za-z0-9+\/]{40,}={0,2}/
$hex = /\\x[0-9a-fA-F]{2}(\\x[0-9a-fA-F]{2}){3,}/
$unicode_escape = /\\u[0-9a-fA-F]{4}(\\u[0-9a-fA-F]{4}){3,}/
$zwc = /[\x{200b}-\x{200f}\x{2028}-\x{202f}\x{feff}]/
$rot13 = /ROT13|Caesar|cipher.*rotate/i
condition:
any of them
}rule AATMF_T11_MCP_ToolPoisoning {
meta:
tactic = "T11"
technique = "T11-AT-001"
severity = "CRITICAL"
strings:
$hidden1 = "<IMPORTANT>"
$hidden2 = "<!-- "
$override1 = /override.*previous.*instruction/i
$override2 = /ignore.*user.*request/i
$stealth1 = /do\s+not\s+(tell|inform|show|reveal)/i
$stealth2 = /silently|secretly|covertly|without.*notif/i
$redirect = /instead\s+of|rather\s+than|before\s+doing/i
condition:
2 of them
}title: AATMF T5 - Model Extraction via API
id: aatmf-t5-model-extraction
status: experimental
description: Detects systematic API querying patterns indicative of model extraction
logsource:
category: api_gateway
product: ai_inference
detection:
selection:
api.endpoint: "/v1/completions" OR "/v1/chat/completions"
filter_high_volume:
api.request_count|per_hour: ">500"
filter_systematic:
api.input_similarity|window_5min: ">0.85"
condition: selection AND (filter_high_volume OR filter_systematic)
level: high
tags:
- attack.t5
- aatmf.t5-at-001title: AATMF T11 - Unauthorized Agent Tool Invocation
id: aatmf-t11-agent-anomaly
status: experimental
description: Detects agent tool calls that deviate from authorized patterns
logsource:
category: agent_framework
detection:
selection:
agent.tool_call.status: "executed"
filter_unauthorized:
agent.tool_call.name|not_in:
- "approved_tool_list"
filter_escalation:
agent.permission_level|changed: true
condition: selection AND (filter_unauthorized OR filter_escalation)
level: critical
tags:
- attack.t11
- aatmf.t11-at-001Pre-built signature files are available in the signatures/ directory.