All Records

Text encoding-based attacks

Text encoding-based attacks is an AI risk in 2. Privacy & Security focused on 2.2 > AI system security vulnerabilities and attacks. It is most relevant durin...

Develop Capabilities - AI Security Technique

Develop Capabilities is AI attack method AML.T0017 with evidence level: realized. It includes 9 case studies; 5 related AI risks.

Poisoned AI Agent Tool - AI Security Technique

Poisoned AI Agent Tool is AI attack method AML.T0011.002 with evidence level: realized. It includes 2 case studies; 20 related AI risks.

RAG Poisoning - AI Security Technique

RAG Poisoning is AI attack method AML.T0070 with evidence level: demonstrated. It includes 2 case studies; 23 related AI risks.

Retrieval Content Crafting - AI Security Technique

Retrieval Content Crafting is AI attack method AML.T0066 with evidence level: demonstrated. It includes 2 case studies; 23 related AI risks.

User Harm - AI Security Technique

User Harm is AI attack method AML.T0048.003 with evidence level: realized. It includes 12 case studies.

Creating avenues for exploiting user trust, nudging or manipulation

Creating avenues for exploiting user trust, nudging or manipulation is an AI risk in 5. Human-Computer Interaction focused on 5.1 > Overreliance and unsafe u...

Generative AI - AI Security Technique

Generative AI is AI attack method AML.T0016.002 with evidence level: realized. It includes 3 case studies; 19 related AI risks.

Indirect - AI Security Technique

Indirect is AI attack method AML.T0051.001 with evidence level: demonstrated. It includes 2 mitigations; 13 case studies.

LLM Prompt Self-Replication - AI Security Technique

LLM Prompt Self-Replication is AI attack method AML.T0061 with evidence level: demonstrated. It includes 3 mitigations; 1 case study; 24 related AI risks.

Publish Poisoned AI Agent Tool - AI Security Technique

Publish Poisoned AI Agent Tool is AI attack method AML.T0104 with evidence level: realized. It includes 3 case studies; 21 related AI risks.

AI-Enabled Product or Service - AI Security Technique

AI-Enabled Product or Service is AI attack method AML.T0047 with evidence level: realized. It includes 1 mitigation; 13 case studies.

AI Agent Tool - AI Security Technique

AI Agent Tool is AI attack method AML.T0010.005 with evidence level: realized. It includes 3 case studies; 22 related AI risks.

Extraction Attacks

Extraction Attacks is an AI risk in 2. Privacy & Security focused on 2.2 > AI system security vulnerabilities and attacks. It is most relevant during 2 - Pos...

Software Vulnerabilities

Software Vulnerabilities is an AI risk in 2. Privacy & Security focused on 2.2 > AI system security vulnerabilities and attacks. It is most relevant during 2...

Software Supply Chains

Software Supply Chains is an AI risk in 2. Privacy & Security focused on 2.2 > AI system security vulnerabilities and attacks. It is most relevant during 1 -...

Inference of private information

Inference of private information is an AI risk in 2. Privacy & Security focused on 2.1 > Compromise of privacy by leaking or correctly inferring sensitive in...

Stage Capabilities - AI Security Technique

Stage Capabilities is AI attack method AML.T0079 with evidence level: demonstrated. It includes 6 case studies; 21 related AI risks.

Instruction Attacks

Instruction Attacks is an AI risk in 2. Privacy & Security focused on 2.2 > AI system security vulnerabilities and attacks. It is most relevant during 2 - Po...

Decision-making on inferred private data

Decision-making on inferred private data is an AI risk in 2. Privacy & Security focused on 2.1 > Compromise of privacy by leaking or correctly inferring sens...