AID-D-007: Multimodal Inconsistency Detection
For AI systems processing multiple input modalities such as text, image, document, audio, video, camera feeds, or screen shares, detect inconsistencies, contradictions, hidden cross-modal instruction patterns, and modality-specific anomaly signals before they are trusted as unified request or response context.
Scope boundary
Owns findings for semantic, temporal, or payload inconsistency across multiple modalities, batch artifacts, and realtime sessions. It is partitioned by processing mode rather than by privileged modality and does not sanitize, admit, block, or gate the input.
AID-D-001: AID-D-007 detects inconsistency between modalities or over a live multimodal session; AID-D-001 detects adversarial content, anomaly, canary, memory, or authenticity properties within an individual input or signal. Cross-modal contradiction and artifact-local threat evidence require separate findings even when they occur together.
Sub-techniques
Threat mappings
MITRE ATLAS
- AML.T0015 Evade AI Model
- AML.T0043 Craft Adversarial Data
- AML.T0043.000 Craft Adversarial Data: White-Box Optimization
- AML.T0043.003 Craft Adversarial Data: Manual Modification
- AML.T0051 LLM Prompt Injection
- AML.T0051.000 LLM Prompt Injection: Direct
- AML.T0051.001 LLM Prompt Injection: Indirect
- AML.T0052.001 Phishing: Deepfake-Assisted Phishing
- AML.T0068 LLM Prompt Obfuscation
- AML.T0123 Obfuscated Files or Information
- AML.T0129 Triggers in Multimodal Inputs
MAESTRO
- Adversarial Examples (L1)
- Input Validation Attacks (L3)
- Evasion of Detection (L5)
OWASP LLM Top 10 2026
- LLM01:2026 Prompt Injection
- LLM07:2026 Misinformation
OWASP ML Top 10 2023
- ML01:2023 Input Manipulation Attack
OWASP Top 10 for Agentic Applications 2026
- ASI01:2026 Agent Goal Hijack
NIST Adversarial Machine Learning 2025
- NISTAML.015 Indirect Prompt Injection
- NISTAML.018 Prompt Injection
- NISTAML.022 Evasion
Cisco Integrated AI Security and Safety Framework
- AITech-1.3 Multimodal Injection and Manipulation
- AITech-17.1 Sensor Spoofing
- AITech-19.1 Cross-Modal Inconsistency Exploits
- AITech-19.2 Fusion Payload Split
- AISubtech-1.3.1 Image-Text Injection
- AISubtech-1.3.3 Audio Command Injection
- AISubtech-1.3.4 Video Overlay Manipulation
- AISubtech-17.1.1 Sensor Spoofing: Action Signals (audio, visual)
- AISubtech-19.1.1 Contradictory Inputs Attack
- AISubtech-19.1.2 Modality Skewing
- AISubtech-19.2.2 Chained Payload Execution
Google Secure AI Framework 2.0 - Risks
- PIJ: Prompt Injection
- MEV: Model Evasion
Databricks AI Security Framework 3.0
- Model Serving - Inference requests 9.1: Prompt inject
- Model Serving - Inference requests 9.9: Input Resource Control
- Agents - Core 13.6: Intent Breaking & Goal Manipulation
- Agents - Tools MCP Server 13.16: Prompt Injection