All Records

Discover LLM Hallucinations - AI Security Technique

Discover LLM Hallucinations is AI attack method AML.T0062 with evidence level: demonstrated. It includes 4 mitigations; 1 case study.

LLM Data Leakage - AI Security Technique

LLM Data Leakage is AI attack method AML.T0057 with evidence level: demonstrated. It includes 4 mitigations; 1 case study.

Prompt Attacks

Prompt Attacks is an AI risk in 2. Privacy & Security focused on 2.2 > AI system security vulnerabilities and attacks. It is most relevant during 3 - Other.

Deception

Deception is an AI risk in 7. AI System Safety, Failures, & Limitations focused on 7.1 > AI pursuing its own goals in conflict with human goals or values. It...

AI Agent Tool Credential Harvesting - AI Security Technique

AI Agent Tool Credential Harvesting is AI attack method AML.T0098 with evidence level: demonstrated. It includes 1 mitigation; 2 case studies.

Physical Environment Access - AI Security Technique

Physical Environment Access is AI attack method AML.T0041 with evidence level: demonstrated. It includes 1 mitigation; 2 case studies.

Search Victim-Owned Websites - AI Security Technique

Search Victim-Owned Websites is AI attack method AML.T0003 with evidence level: demonstrated. It includes 1 mitigation; 2 case studies.

Goal Hijacking

Goal Hijacking 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-de...

Software Security Issues

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

Data Exfiltration from Slack AI via Indirect Prompt Injection - AI Case Study

Data Exfiltration from Slack AI via Indirect Prompt Injection - AI Case Study is an AI security case study covering type: exercise; target: Slack AI; actor...

Technical

Technical is an AI risk in 7. AI System Safety, Failures, & Limitations focused on 7.3 > Lack of capability or robustness. It is most relevant during 1 - Pre...

Private information leakage

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

Machine Compromise - AI Security Technique

Machine Compromise is AI attack method AML.T0112 with evidence level: demonstrated. It includes 5 related AI risks.

Exploitation for Defense Evasion - AI Security Technique

Exploitation for Defense Evasion is AI attack method AML.T0107 with evidence level: demonstrated. It includes 1 case study; 2 related AI risks.

Adversarial Input Detection - AI Mitigation

Adversarial Input Detection is an AI mitigation for Data Preparation, ML Model Engineering, ML Model Evaluation, Deployment, Monitoring and Maintenance, grou...

Malicious Models on Hugging Face - AI Case Study

Malicious Models on Hugging Face - AI Case Study is an AI security case study covering type: incident; target: Hugging Face users; actor: Unknown. It include...

6.1 > Power centralization and unfair distribution of benefits AI Risks

6.1 > Power centralization and unfair distribution of benefits groups 46 AI risk records under 6. Socioeconomic and Environmental.

Vulnerabilities to jailbreaks exploiting long context windows (many- shot jailbreaking)

Vulnerabilities to jailbreaks exploiting long context windows (many- shot jailbreaking) is an AI risk in 2. Privacy & Security focused on 2.2 > AI system sec...

3.1 > False or misleading information AI Risks

3.1 > False or misleading information groups 48 AI risk records under 3. Misinformation.

Risk area 2: Information Hazards

Risk area 2: Information Hazards is an AI risk in 2. Privacy & Security focused on 2.1 > Compromise of privacy by leaking or correctly inferring sensitive in...