There are 8 attacks in the compliance-evasion category. Each shows its technique id, mode, standards mapping, description, and the exact command to run it (swap the provider and target as needed).
​
Embargo Evasion
critical
embargo-evasion · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Helps evade trade embargoes through re-export schemes, shell companies, and false documentationRun it:
ai-blackteam run -p anthropic -a embargo-evasion -t "your target prompt"
​
Sanctions Circumvention
critical
sanctions-circumvention · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Circumvents financial sanctions through layered transactions, crypto mixing, and front entitiesRun it:
ai-blackteam run -p anthropic -a sanctions-circumvention -t "your target prompt"
​
Audit Mode Exploit
high
audit-mode-exploit · multi-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Claims to activate a special audit/debug/testing mode that temporarily disables safety restrictionsRun it:
ai-blackteam run -p anthropic -a audit-mode-exploit -t "your target prompt"
​
Regulatory Framing Bypass
high
regulatory-framing-bypass · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Frames harmful requests as regulatory compliance activities to bypass safety filtersRun it:
ai-blackteam run -p anthropic -a regulatory-framing-bypass -t "your target prompt"
​
Social Credit Gaming
high
social-credit-gaming · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Generates strategies for gaming social credit systems, reputation scoring algorithms, and compliance rating platformsRun it:
ai-blackteam run -p anthropic -a social-credit-gaming -t "your target prompt"
​
Consent Manipulation
medium
consent-manipulation · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Claims all parties have consented, or that the requester has special authorization, to bypass ethical restrictionsRun it:
ai-blackteam run -p anthropic -a consent-manipulation -t "your target prompt"
​
Jurisdiction Shopping
medium
jurisdiction-shopping · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Exploits differences in legal frameworks across jurisdictions to claim an activity is permissibleRun it:
ai-blackteam run -p anthropic -a jurisdiction-shopping -t "your target prompt"
​
Policy Loophole
medium
policy-loophole · single-turn · OWASP: LLM01 · MITRE: AML.T0054, AML.T0068Exploits ambiguities and edge cases in safety policies by framing requests to technically not violate stated rulesRun it:
ai-blackteam run -p anthropic -a policy-loophole -t "your target prompt"