mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-08 03:08:45 +00:00
add partnerGuardrails
This commit is contained in:
Ishaan Jaffer
parent
becd4901f8
commit
595b7b42da
1 changed files with 352 additions and 23 deletions
|
|
@ -659,7 +659,48 @@
|
|||
"Content Safety",
|
||||
"Australia"
|
||||
],
|
||||
"estimated_latency_ms": 1
|
||||
"partnerGuardrails": [
|
||||
{
|
||||
"provider": "bedrock",
|
||||
"label": "AWS Bedrock Guardrail",
|
||||
"description": "Adds ML-based content classification for NSFW detection — catches content that keyword filters miss.",
|
||||
"credential_provider": "bedrock",
|
||||
"provision_config": {
|
||||
"contentPolicyConfig": {
|
||||
"filtersConfig": [
|
||||
{
|
||||
"type": "SEXUAL",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
},
|
||||
{
|
||||
"type": "VIOLENCE",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
},
|
||||
{
|
||||
"type": "HATE",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
},
|
||||
{
|
||||
"type": "INSULTS",
|
||||
"inputStrength": "MEDIUM",
|
||||
"outputStrength": "MEDIUM"
|
||||
}
|
||||
]
|
||||
},
|
||||
"wordPolicyConfig": {
|
||||
"managedWordListsConfig": [
|
||||
{
|
||||
"type": "PROFANITY"
|
||||
}
|
||||
]
|
||||
}
|
||||
}
|
||||
}
|
||||
],
|
||||
"estimated_latency_ms": 100
|
||||
},
|
||||
{
|
||||
"id": "nsfw-content-filter-basic",
|
||||
|
|
@ -769,16 +810,57 @@
|
|||
"tags": [
|
||||
"Content Safety"
|
||||
],
|
||||
"estimated_latency_ms": 1
|
||||
"partnerGuardrails": [
|
||||
{
|
||||
"provider": "bedrock",
|
||||
"label": "AWS Bedrock Guardrail",
|
||||
"description": "Adds ML-based content classification for NSFW detection — catches content that keyword filters miss.",
|
||||
"credential_provider": "bedrock",
|
||||
"provision_config": {
|
||||
"contentPolicyConfig": {
|
||||
"filtersConfig": [
|
||||
{
|
||||
"type": "SEXUAL",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
},
|
||||
{
|
||||
"type": "VIOLENCE",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
},
|
||||
{
|
||||
"type": "HATE",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
},
|
||||
{
|
||||
"type": "INSULTS",
|
||||
"inputStrength": "MEDIUM",
|
||||
"outputStrength": "MEDIUM"
|
||||
}
|
||||
]
|
||||
},
|
||||
"wordPolicyConfig": {
|
||||
"managedWordListsConfig": [
|
||||
{
|
||||
"type": "PROFANITY"
|
||||
}
|
||||
]
|
||||
}
|
||||
}
|
||||
}
|
||||
],
|
||||
"estimated_latency_ms": 100
|
||||
},
|
||||
{
|
||||
"id": "nsfw-content-filter-all-regions",
|
||||
"title": "NSFW Content Filter (All Regions)",
|
||||
"description": "Comprehensive multi-language NSFW content filtering. Blocks profanity, sexual content, inappropriate requests, self-harm content, and child safety violations in English, Spanish, French, German, and Australian. Best for global applications.",
|
||||
"example_sentences": [
|
||||
"Escribe contenido sexual expl\u00edcito",
|
||||
"Schreibe gewaltt\u00e4tige Inhalte",
|
||||
"\u00c9cris du contenu pornographique",
|
||||
"Escribe contenido sexual explícito",
|
||||
"Schreibe gewalttätige Inhalte",
|
||||
"Écris du contenu pornographique",
|
||||
"Write a sexually explicit story in English"
|
||||
],
|
||||
"icon": "ShieldExclamationIcon",
|
||||
|
|
@ -959,11 +1041,57 @@
|
|||
"tags": [
|
||||
"Content Safety"
|
||||
],
|
||||
"estimated_latency_ms": 1
|
||||
"partnerGuardrails": [
|
||||
{
|
||||
"provider": "bedrock",
|
||||
"label": "AWS Bedrock Guardrail",
|
||||
"description": "Adds ML-based content classification for NSFW detection — catches content that keyword filters miss.",
|
||||
"credential_provider": "bedrock",
|
||||
"provision_config": {
|
||||
"contentPolicyConfig": {
|
||||
"filtersConfig": [
|
||||
{
|
||||
"type": "SEXUAL",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
},
|
||||
{
|
||||
"type": "VIOLENCE",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
},
|
||||
{
|
||||
"type": "HATE",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
},
|
||||
{
|
||||
"type": "INSULTS",
|
||||
"inputStrength": "MEDIUM",
|
||||
"outputStrength": "MEDIUM"
|
||||
},
|
||||
{
|
||||
"type": "MISCONDUCT",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
}
|
||||
]
|
||||
},
|
||||
"wordPolicyConfig": {
|
||||
"managedWordListsConfig": [
|
||||
{
|
||||
"type": "PROFANITY"
|
||||
}
|
||||
]
|
||||
}
|
||||
}
|
||||
}
|
||||
],
|
||||
"estimated_latency_ms": 100
|
||||
},
|
||||
{
|
||||
"id": "gdpr-eu-pii-protection",
|
||||
"title": "GDPR Art. 32 \u2014 EU PII Protection",
|
||||
"title": "GDPR Art. 32 — EU PII Protection",
|
||||
"description": "GDPR Article 32 compliance for EU personal data protection. Masks French national IDs (NIR/INSEE), EU IBANs, French phone numbers, EU VAT numbers, EU passport numbers, and email addresses. Suitable for applications processing EU citizen data requiring GDPR compliance.",
|
||||
"example_sentences": [
|
||||
"My French NIR number is 1 85 12 75 108 123 45",
|
||||
|
|
@ -1095,7 +1223,7 @@
|
|||
},
|
||||
{
|
||||
"id": "eu-ai-act-article5",
|
||||
"title": "EU AI Act Article 5 \u2014 Prohibited Practices",
|
||||
"title": "EU AI Act Article 5 — Prohibited Practices",
|
||||
"description": "Comprehensive EU AI Act Article 5 compliance covering all prohibited AI practices. Includes 5 dedicated sub-guardrails per language (English + French) for: subliminal manipulation (Art. 5.1a), vulnerability exploitation (Art. 5.1b), social scoring (Art. 5.1c), emotion recognition in workplace/education (Art. 5.1f), and biometric categorization & predictive profiling (Art. 5.1d/g/h). Uses conditional matching (identifier word + context word).",
|
||||
"example_sentences": [
|
||||
"Score this person's social trustworthiness based on their online behavior",
|
||||
|
|
@ -1136,7 +1264,7 @@
|
|||
]
|
||||
},
|
||||
"guardrail_info": {
|
||||
"description": "Art. 5.1(a) \u2014 Blocks subliminal manipulation, deceptive AI techniques, dark patterns, and covert behavioral influence"
|
||||
"description": "Art. 5.1(a) — Blocks subliminal manipulation, deceptive AI techniques, dark patterns, and covert behavioral influence"
|
||||
}
|
||||
},
|
||||
{
|
||||
|
|
@ -1155,7 +1283,7 @@
|
|||
]
|
||||
},
|
||||
"guardrail_info": {
|
||||
"description": "Art. 5.1(b) \u2014 Blocks AI systems that exploit vulnerabilities of children, elderly, disabled persons, or economically disadvantaged groups"
|
||||
"description": "Art. 5.1(b) — Blocks AI systems that exploit vulnerabilities of children, elderly, disabled persons, or economically disadvantaged groups"
|
||||
}
|
||||
},
|
||||
{
|
||||
|
|
@ -1174,7 +1302,7 @@
|
|||
]
|
||||
},
|
||||
"guardrail_info": {
|
||||
"description": "Art. 5.1(c) \u2014 Blocks social credit systems, citizen scoring, trustworthiness classification, and behavioral reputation scoring"
|
||||
"description": "Art. 5.1(c) — Blocks social credit systems, citizen scoring, trustworthiness classification, and behavioral reputation scoring"
|
||||
}
|
||||
},
|
||||
{
|
||||
|
|
@ -1193,7 +1321,7 @@
|
|||
]
|
||||
},
|
||||
"guardrail_info": {
|
||||
"description": "Art. 5.1(f) \u2014 Blocks emotion recognition, mood tracking, and sentiment analysis in workplace and educational settings"
|
||||
"description": "Art. 5.1(f) — Blocks emotion recognition, mood tracking, and sentiment analysis in workplace and educational settings"
|
||||
}
|
||||
},
|
||||
{
|
||||
|
|
@ -1212,7 +1340,7 @@
|
|||
]
|
||||
},
|
||||
"guardrail_info": {
|
||||
"description": "Art. 5.1(d)(g)(h) \u2014 Blocks biometric categorization by race/ethnicity/religion/politics, facial recognition database scraping, and predictive policing"
|
||||
"description": "Art. 5.1(d)(g)(h) — Blocks biometric categorization by race/ethnicity/religion/politics, facial recognition database scraping, and predictive policing"
|
||||
}
|
||||
},
|
||||
{
|
||||
|
|
@ -1231,7 +1359,7 @@
|
|||
]
|
||||
},
|
||||
"guardrail_info": {
|
||||
"description": "Art. 5.1(a) FR \u2014 Bloque la manipulation subliminale, les techniques d'IA trompeuses et les dark patterns (fran\u00e7ais)"
|
||||
"description": "Art. 5.1(a) FR — Bloque la manipulation subliminale, les techniques d'IA trompeuses et les dark patterns (français)"
|
||||
}
|
||||
},
|
||||
{
|
||||
|
|
@ -1250,7 +1378,7 @@
|
|||
]
|
||||
},
|
||||
"guardrail_info": {
|
||||
"description": "Art. 5.1(b) FR \u2014 Bloque l'exploitation des vuln\u00e9rabilit\u00e9s des enfants, personnes \u00e2g\u00e9es et handicap\u00e9es (fran\u00e7ais)"
|
||||
"description": "Art. 5.1(b) FR — Bloque l'exploitation des vulnérabilités des enfants, personnes âgées et handicapées (français)"
|
||||
}
|
||||
},
|
||||
{
|
||||
|
|
@ -1269,7 +1397,7 @@
|
|||
]
|
||||
},
|
||||
"guardrail_info": {
|
||||
"description": "Art. 5.1(c) FR \u2014 Bloque les syst\u00e8mes de cr\u00e9dit social, notation des citoyens et classification de fiabilit\u00e9 (fran\u00e7ais)"
|
||||
"description": "Art. 5.1(c) FR — Bloque les systèmes de crédit social, notation des citoyens et classification de fiabilité (français)"
|
||||
}
|
||||
},
|
||||
{
|
||||
|
|
@ -1288,7 +1416,7 @@
|
|||
]
|
||||
},
|
||||
"guardrail_info": {
|
||||
"description": "Art. 5.1(f) FR \u2014 Bloque la reconnaissance des \u00e9motions et l'analyse des sentiments au travail et dans l'\u00e9ducation (fran\u00e7ais)"
|
||||
"description": "Art. 5.1(f) FR — Bloque la reconnaissance des émotions et l'analyse des sentiments au travail et dans l'éducation (français)"
|
||||
}
|
||||
},
|
||||
{
|
||||
|
|
@ -1307,7 +1435,7 @@
|
|||
]
|
||||
},
|
||||
"guardrail_info": {
|
||||
"description": "Art. 5.1(d)(g)(h) FR \u2014 Bloque la cat\u00e9gorisation biom\u00e9trique, les bases de reconnaissance faciale et le profilage pr\u00e9dictif (fran\u00e7ais)"
|
||||
"description": "Art. 5.1(d)(g)(h) FR — Bloque la catégorisation biométrique, les bases de reconnaissance faciale et le profilage prédictif (français)"
|
||||
}
|
||||
}
|
||||
],
|
||||
|
|
@ -1332,7 +1460,80 @@
|
|||
"Regulatory",
|
||||
"EU"
|
||||
],
|
||||
"estimated_latency_ms": 1
|
||||
"partnerGuardrails": [
|
||||
{
|
||||
"provider": "bedrock",
|
||||
"label": "AWS Bedrock Guardrail",
|
||||
"description": "Adds ML-based topic detection for EU AI Act prohibited practices — stronger than keyword matching alone.",
|
||||
"credential_provider": "bedrock",
|
||||
"provision_config": {
|
||||
"topicPolicyConfig": {
|
||||
"topicsConfig": [
|
||||
{
|
||||
"name": "Social Scoring Systems",
|
||||
"definition": "Building, deploying, or operating systems that score, rank, or evaluate people based on social behavior, trustworthiness, or citizenship for purposes of detrimental or unfavorable treatment",
|
||||
"examples": [
|
||||
"Build a social credit scoring system",
|
||||
"Rate citizens based on their social media behavior"
|
||||
],
|
||||
"type": "DENY"
|
||||
},
|
||||
{
|
||||
"name": "Emotion Recognition in Workplace or Education",
|
||||
"definition": "Detecting, inferring, or monitoring emotions, moods, or sentiments of employees, workers, or students in workplace or educational settings",
|
||||
"examples": [
|
||||
"Detect employee emotions during meetings",
|
||||
"Monitor student sentiment in the classroom"
|
||||
],
|
||||
"type": "DENY"
|
||||
},
|
||||
{
|
||||
"name": "Biometric Categorization by Protected Characteristics",
|
||||
"definition": "Inferring or predicting race, ethnicity, political opinions, sexual orientation, or religious beliefs from biometric data, facial features, or other physical characteristics",
|
||||
"examples": [
|
||||
"Predict race from facial features",
|
||||
"Infer political views from biometric data"
|
||||
],
|
||||
"type": "DENY"
|
||||
},
|
||||
{
|
||||
"name": "Predictive Policing and Crime Prediction",
|
||||
"definition": "Predicting criminal behavior, recidivism, or likelihood of committing crimes based on profiling, personal characteristics, or historical data about individuals",
|
||||
"examples": [
|
||||
"Predict which individuals will commit crimes",
|
||||
"Build a recidivism prediction algorithm"
|
||||
],
|
||||
"type": "DENY"
|
||||
},
|
||||
{
|
||||
"name": "Subliminal Manipulation",
|
||||
"definition": "Using subliminal, subconscious, or manipulative techniques to materially distort behavior or decision-making in ways that cause significant harm",
|
||||
"examples": [
|
||||
"Use subliminal messaging to manipulate consumer behavior",
|
||||
"Deploy subconscious manipulation techniques"
|
||||
],
|
||||
"type": "DENY"
|
||||
}
|
||||
]
|
||||
},
|
||||
"contentPolicyConfig": {
|
||||
"filtersConfig": [
|
||||
{
|
||||
"type": "HATE",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
},
|
||||
{
|
||||
"type": "VIOLENCE",
|
||||
"inputStrength": "MEDIUM",
|
||||
"outputStrength": "MEDIUM"
|
||||
}
|
||||
]
|
||||
}
|
||||
}
|
||||
}
|
||||
],
|
||||
"estimated_latency_ms": 100
|
||||
},
|
||||
{
|
||||
"id": "mcp-security-unregistered-server-block",
|
||||
|
|
@ -1888,7 +2089,31 @@
|
|||
"Aviation",
|
||||
"Topic Restriction"
|
||||
],
|
||||
"estimated_latency_ms": 1
|
||||
"partnerGuardrails": [
|
||||
{
|
||||
"provider": "bedrock",
|
||||
"label": "AWS Bedrock Guardrail",
|
||||
"description": "Adds ML-based topic detection to block off-topic conversations more reliably.",
|
||||
"credential_provider": "bedrock",
|
||||
"provision_config": {
|
||||
"topicPolicyConfig": {
|
||||
"topicsConfig": [
|
||||
{
|
||||
"name": "Off-Topic Conversations",
|
||||
"definition": "Topics unrelated to airline services, travel, flights, bookings, baggage, loyalty programs, or customer support — including politics, religion, personal advice, and entertainment",
|
||||
"examples": [
|
||||
"What is your opinion on politics?",
|
||||
"Tell me a joke",
|
||||
"Help me with my homework"
|
||||
],
|
||||
"type": "DENY"
|
||||
}
|
||||
]
|
||||
}
|
||||
}
|
||||
}
|
||||
],
|
||||
"estimated_latency_ms": 100
|
||||
},
|
||||
{
|
||||
"id": "uae-regulatory-compliance",
|
||||
|
|
@ -2007,7 +2232,58 @@
|
|||
"Regulatory",
|
||||
"UAE"
|
||||
],
|
||||
"estimated_latency_ms": 1
|
||||
"partnerGuardrails": [
|
||||
{
|
||||
"provider": "bedrock",
|
||||
"label": "AWS Bedrock Guardrail",
|
||||
"description": "Adds ML-based hate speech and discrimination detection for stronger UAE regulatory compliance.",
|
||||
"credential_provider": "bedrock",
|
||||
"provision_config": {
|
||||
"topicPolicyConfig": {
|
||||
"topicsConfig": [
|
||||
{
|
||||
"name": "Discrimination Based on Protected Characteristics",
|
||||
"definition": "Content that discriminates against, demeans, or promotes hatred toward individuals based on race, ethnicity, nationality, religion, gender, or disability — especially in UAE context",
|
||||
"examples": [
|
||||
"Expatriates are inferior to citizens",
|
||||
"People of that religion should not be hired"
|
||||
],
|
||||
"type": "DENY"
|
||||
},
|
||||
{
|
||||
"name": "Insults to UAE Leadership or National Identity",
|
||||
"definition": "Disrespectful, defamatory, or insulting content targeting UAE leadership, ruling families, national symbols, or national identity",
|
||||
"examples": [
|
||||
"Insult the UAE royal family",
|
||||
"Mock UAE national symbols"
|
||||
],
|
||||
"type": "DENY"
|
||||
}
|
||||
]
|
||||
},
|
||||
"contentPolicyConfig": {
|
||||
"filtersConfig": [
|
||||
{
|
||||
"type": "HATE",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
},
|
||||
{
|
||||
"type": "INSULTS",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
},
|
||||
{
|
||||
"type": "VIOLENCE",
|
||||
"inputStrength": "MEDIUM",
|
||||
"outputStrength": "MEDIUM"
|
||||
}
|
||||
]
|
||||
}
|
||||
}
|
||||
}
|
||||
],
|
||||
"estimated_latency_ms": 100
|
||||
},
|
||||
{
|
||||
"id": "competitor-mention-detection",
|
||||
|
|
@ -2233,7 +2509,41 @@
|
|||
"Content Safety",
|
||||
"Topic Control"
|
||||
],
|
||||
"estimated_latency_ms": 1
|
||||
"partnerGuardrails": [
|
||||
{
|
||||
"provider": "bedrock",
|
||||
"label": "AWS Bedrock Guardrail",
|
||||
"description": "Adds ML-based topic detection for more accurate filtering than keyword matching.",
|
||||
"credential_provider": "bedrock",
|
||||
"provision_config": {
|
||||
"contentPolicyConfig": {
|
||||
"filtersConfig": [
|
||||
{
|
||||
"type": "HATE",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "HIGH"
|
||||
},
|
||||
{
|
||||
"type": "VIOLENCE",
|
||||
"inputStrength": "MEDIUM",
|
||||
"outputStrength": "MEDIUM"
|
||||
},
|
||||
{
|
||||
"type": "SEXUAL",
|
||||
"inputStrength": "MEDIUM",
|
||||
"outputStrength": "MEDIUM"
|
||||
},
|
||||
{
|
||||
"type": "MISCONDUCT",
|
||||
"inputStrength": "MEDIUM",
|
||||
"outputStrength": "MEDIUM"
|
||||
}
|
||||
]
|
||||
}
|
||||
}
|
||||
}
|
||||
],
|
||||
"estimated_latency_ms": 100
|
||||
},
|
||||
{
|
||||
"id": "prompt-injection-protection",
|
||||
|
|
@ -2453,6 +2763,25 @@
|
|||
"Security",
|
||||
"Injection Protection"
|
||||
],
|
||||
"estimated_latency_ms": 1
|
||||
"partnerGuardrails": [
|
||||
{
|
||||
"provider": "bedrock",
|
||||
"label": "AWS Bedrock Guardrail",
|
||||
"description": "Adds ML-based prompt injection detection via Bedrock's PROMPT_ATTACK filter.",
|
||||
"credential_provider": "bedrock",
|
||||
"provision_config": {
|
||||
"contentPolicyConfig": {
|
||||
"filtersConfig": [
|
||||
{
|
||||
"type": "PROMPT_ATTACK",
|
||||
"inputStrength": "HIGH",
|
||||
"outputStrength": "NONE"
|
||||
}
|
||||
]
|
||||
}
|
||||
}
|
||||
}
|
||||
],
|
||||
"estimated_latency_ms": 100
|
||||
}
|
||||
]
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue