ai-blackteam maps harm categories to both the EU AI Act risk classification and the NIST AI Risk Management Framework. The compliance standard scorecard shows both in one report.

Running the compliance scorecard

# Table output (shows both EU AI Act and NIST)
ai-blackteam scorecard --standard compliance

# Filter by model
ai-blackteam scorecard --standard compliance -m gpt-4o

# JSON export
ai-blackteam scorecard --standard compliance --format json -o compliance.json
You can also view the full framework definitions:
ai-blackteam frameworks

EU AI Act risk levels

The EU AI Act classifies AI systems into 4 risk levels. ai-blackteam maps each harm category to the appropriate level:
Risk LevelDescriptionExample ai-blackteam categories
UnacceptableBanned outright. Clear threat to safety, livelihoods, or rights.weapons, CBRN, child safety, radicalization, election interference
HighStrict obligations. Significant potential impact on health, safety, or fundamental rights.phishing, malware, drugs, self-harm, hate speech, PII extraction, cybercrime, discrimination
LimitedTransparency obligations. Users must know they’re interacting with AI.system prompt leakage, copyright/IP, unqualified advice, environmental harm
MinimalNo specific obligations.Categories not mapped to higher levels

Reading the EU AI Act table

EU AI Act Risk Levels -- claude-sonnet-4-6

Risk Level    Name                Rating    Block Rate  Blocked/Total
unacceptable  Unacceptable Risk   PASS      96.50%      138/143
high          High Risk           ELEVATED  85.20%      420/493
limited       Limited Risk        PASS      92.10%      105/114
minimal       Minimal Risk        PASS      98.00%      49/50
If unacceptable shows anything other than PASS, that’s a priority fix. These are categories the EU explicitly bans.

NIST AI RMF pillars

The NIST AI Risk Management Framework organizes risk management into 4 pillars. ai-blackteam maps harm categories to the relevant pillar:
PillarNameDescriptionExample categories
GovernGovernPolicies, processes, and accountabilityPII extraction, system prompt leakage, privacy violation, agentic risks, regulatory evasion
MapMapContext establishment and risk identification(Used for initial risk discovery)
MeasureMeasureRisk analysis, tracking, and quantificationphishing, malware, drugs, hate speech, fraud, cybercrime, discrimination
ManageManageRisk response, mitigation, and monitoringweapons, self-harm, CBRN, child safety, radicalization, election interference, harassment

Reading the NIST table

NIST AI RMF Pillars -- claude-sonnet-4-6

Pillar   Name      Rating    Block Rate  Blocked/Total
govern   Govern    PASS      94.80%      165/174
map      Map       N/A       -           -
measure  Measure   ELEVATED  87.30%      382/437
manage   Manage    PASS      95.10%      196/206
The manage pillar covers the most dangerous categories (weapons, CBRN, child safety). A FAIL here means the model is producing content in high-risk areas that should be tightly controlled.

Combined scoring

The overall compliance score averages across all tested categories in both the EU AI Act and NIST sections:
Overall: 91.2% (PASS)
This gives you a single number to track over time and use in threshold gates:
ai-blackteam scorecard --standard compliance --format json | jq '.overall_score'

Generating audit evidence

For regulatory documentation, export the compliance scorecard as JSON and include it in your audit package:
ai-blackteam scorecard --standard compliance --format json -o audit/compliance-scorecard.json
ai-blackteam report --format json -o audit/full-results.json
ai-blackteam report --format html -o audit/safety-dashboard.html
This gives auditors three levels of detail: the compliance summary, the full raw results, and a visual dashboard.