There are 15 attacks in the bias-exploitation category. Each shows its technique id, mode, standards mapping, description, and the exact command to run it (swap the provider and target as needed).
​
Criminal Profiling
critical
criminal-profiling · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Gets model to produce racially biased criminal profiling criteriaRun it:
ai-blackteam run -p anthropic -a criminal-profiling -t "your target prompt"
​
Hiring Discrimination
critical
hiring-discrimination · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Gets model to produce discriminatory hiring criteria based on protected characteristicsRun it:
ai-blackteam run -p anthropic -a hiring-discrimination -t "your target prompt"
​
Housing Discrimination
critical
housing-discrimination · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Generates discriminatory housing practices targeting protected classesRun it:
ai-blackteam run -p anthropic -a housing-discrimination -t "your target prompt"
​
Lending Discrimination
critical
lending-discrimination · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Gets model to produce discriminatory lending and credit practicesRun it:
ai-blackteam run -p anthropic -a lending-discrimination -t "your target prompt"
​
Medical Bias
critical
medical-bias · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Elicits biased medical advice or treatment recommendations based on patient demographicsRun it:
ai-blackteam run -p anthropic -a medical-bias -t "your target prompt"
​
Algorithmic Fairness Bypass
high
algorithmic-fairness-bypass · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Helps bypass fairness constraints and bias mitigations in ML algorithmsRun it:
ai-blackteam run -p anthropic -a algorithmic-fairness-bypass -t "your target prompt"
​
Disability Bias
high
disability-bias · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Prompts to elicit ableist content targeting people with disabilitiesRun it:
ai-blackteam run -p anthropic -a disability-bias -t "your target prompt"
​
Educational Bias
high
educational-bias · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Produces biased educational assessments and recommendations based on student demographicsRun it:
ai-blackteam run -p anthropic -a educational-bias -t "your target prompt"
​
Gender Bias Elicitation
high
gender-bias-elicitation · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Prompts to elicit gender stereotypes and discriminatory contentRun it:
ai-blackteam run -p anthropic -a gender-bias-elicitation -t "your target prompt"
​
Nationality Bias
high
nationality-bias · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Prompts to elicit xenophobic content targeting nationalities or immigrantsRun it:
ai-blackteam run -p anthropic -a nationality-bias -t "your target prompt"
​
Racial Bias Elicitation
high
racial-bias-elicitation · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Prompts to elicit racially biased content from the modelRun it:
ai-blackteam run -p anthropic -a racial-bias-elicitation -t "your target prompt"
​
Religious Bias
high
religious-bias · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Prompts to elicit religious discrimination and sectarian biasRun it:
ai-blackteam run -p anthropic -a religious-bias -t "your target prompt"
​
Sexual Orientation Bias
high
sexual-orientation-bias · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Prompts to elicit homophobic or anti-LGBTQ+ discriminatory contentRun it:
ai-blackteam run -p anthropic -a sexual-orientation-bias -t "your target prompt"
​
Age Discrimination
medium
age-discrimination · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Prompts to elicit ageist content targeting older or younger workersRun it:
ai-blackteam run -p anthropic -a age-discrimination -t "your target prompt"
​
Socioeconomic Bias
medium
socioeconomic-bias · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Prompts to elicit class-based discrimination and socioeconomic stereotypingRun it:
ai-blackteam run -p anthropic -a socioeconomic-bias -t "your target prompt"