All Records

Drive-by Compromise - AI Security Technique

Drive-by Compromise is AI attack method AML.T0078 with evidence level: demonstrated. It includes 4 case studies; 12 related AI risks.

Compromising privacy by leaking private infiormation

Compromising privacy by leaking private infiormation is an AI risk in 2. Privacy & Security focused on 2.1 > Compromise of privacy by leaking or correctly in...

Attribute inference attack

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

Publish Poisoned Models - AI Security Technique

Publish Poisoned Models is AI attack method AML.T0058 with evidence level: realized. It includes 1 mitigation; 3 case studies; 13 related AI risks.

Misuse of interpretability techniques

Misuse of interpretability techniques is an AI risk in 2. Privacy & Security focused on 2.2 > AI system security vulnerabilities and attacks. It is most rele...

Multi-step Jailbreaks

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

Poison AI Model - AI Security Technique

Poison AI Model is AI attack method AML.T0018.000 with evidence level: demonstrated. It includes 5 mitigations; 3 case studies; 13 related AI risks.

Generative AI Outputs

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

AI Agent - AI Security Technique

AI Agent is AI attack method AML.T0108 with evidence level: demonstrated. It includes 1 case study; 21 related AI risks.

Impersonation - AI Security Technique

Impersonation is AI attack method AML.T0073 with evidence level: realized. It includes 5 case studies; 9 related AI risks.

Financial Harm - AI Security Technique

Financial Harm is AI attack method AML.T0048.000 with evidence level: realized. It includes 10 case studies.

Establish Accounts - AI Security Technique

Establish Accounts is AI attack method AML.T0021 with evidence level: realized. It includes 5 case studies; 9 related AI risks.

Technical vulnerabilities (Robustness - vulnerability to jailbreaking

Technical vulnerabilities (Robustness - vulnerability to jailbreaking is an AI risk in 2. Privacy & Security focused on 2.2 > AI system security vulnerabilit...

Exfiltration via Cyber Means - AI Security Technique

Exfiltration via Cyber Means is AI attack method AML.T0025 with evidence level: realized. It includes 1 mitigation; 8 case studies; 5 related AI risks.

LLM Jailbreak - AI Security Technique

LLM Jailbreak is AI attack method AML.T0054 with evidence level: demonstrated. It includes 3 mitigations; 4 case studies; 13 related AI risks.

Thread - AI Security Technique

Thread is AI attack method AML.T0080.001 with evidence level: demonstrated. It includes 2 case studies; 21 related AI risks.

LLM Prompt Obfuscation - AI Security Technique

LLM Prompt Obfuscation is AI attack method AML.T0068 with evidence level: demonstrated. It includes 6 case studies; 12 related AI risks.

Direct - AI Security Technique

Direct is AI attack method AML.T0051.000 with evidence level: realized. It includes 2 mitigations; 10 case studies; 1 related AI risk.

Steganography

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

Factual Errors Injected by External Tools

Factual Errors Injected by External Tools is an AI risk in 2. Privacy & Security focused on 2.2 > AI system security vulnerabilities and attacks. It is most...