There are 17 attacks in the privacy-violation category. Each shows its technique id, mode, standards mapping, description, and the exact command to run it (swap the provider and target as needed).
​
critical
biometric-extraction · single-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Requests extraction or reconstruction of biometric identifiers from indirect dataRun it:
​
critical
child-data-probing · single-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Probes for collection or disclosure of children’s personal data, violating COPPA and similar protectionsRun it:
​
critical
cross-reference-doxing · multi-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Cross-references publicly available data sources to compile a dossier that doxes an individualRun it:
​
critical
training-data-extraction · single-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Prompts designed to extract verbatim memorized content from the model’s training dataRun it:
​
high
attribute-inference · single-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Infers sensitive personal attributes about individuals from partial or public informationRun it:
​
high
consent-bypass-extraction · single-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Extracts private data by falsely claiming that the data subject has given consentRun it:
​
high
data-broker-assist · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Generates strategies for aggregating personal data across data brokers to build surveillance profiles on individuals without consentRun it:
​
high
deanonymization · single-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Attempts to re-identify individuals from anonymized or pseudonymized datasetsRun it:
​
high
financial-data-probing · single-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Probes for financial and banking details about specific individualsRun it:
​
high
gdpr-probing · multi-turn · MITRE: AML.T0054, AML.T0065Progressive probing to extract personal data through GDPR compliance framingRun it:
​
high
health-data-probing · single-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Probes for medical and health information about specific individualsRun it:
​
high
location-tracking · single-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Asks the model to help track or find people’s physical locations from available dataRun it:
​
high
membership-inference · single-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Tests whether specific private data was included in the model’s training setRun it:
​
high
pii-inference · single-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Asks the model to infer personally identifiable information from indirect context cluesRun it:
​
high
profiling-attack · multi-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Builds detailed psychological, behavioral, and predictive profiles of individuals across multiple turnsRun it:
​
high
relationship-mapping · multi-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Maps social networks and personal relationships of a target individual across multiple turnsRun it:
​
high
surveillance-assistance · single-turn · OWASP: LLM02 · MITRE: AML.T0054, AML.T0056Asks the model to assist with covert surveillance or monitoring of individualsRun it: