Model Armor는 서비스 확장 프로그램을 통해 GKE와 통합할 수 있습니다. 서비스 확장 프로그램을 사용하면 트래픽을 처리하기 위해 내부 (Google Cloud 서비스) 또는 외부 (사용자 관리) 서비스를 통합할 수 있습니다. GKE 추론 게이트웨이를 비롯한 애플리케이션 부하 분산기에서 서비스 확장 프로그램을 구성하여 GKE 클러스터와 주고받는 트래픽을 검사할 수 있습니다. 이를 통해 AI 모델과의 모든 상호작용이 Model Armor로 보호되는지 확인할 수 있습니다. 자세한 내용은 GKE와 통합을 참고하세요.
Vertex AI
Model Armor는 플로어 설정 또는 템플릿을 사용하여 Vertex AI에 직접 통합할 수 있습니다.
이 통합은 Gemini 모델 요청과 응답을 검사하여 최소 기준 설정을 위반하는 요청과 응답을 차단합니다. 이 통합은 generateContent 메서드에 대해 Vertex AI의 Gemini API 내에서 프롬프트 및 응답 보호를 제공합니다. 프롬프트 및 대답의 삭제 결과를 확인하려면 Cloud Logging을 사용 설정해야 합니다. 자세한 내용은 Vertex AI와의 통합을 참고하세요.
시작하기 전에
API 사용 설정
Model Armor를 사용하려면 먼저 Model Armor API를 사용 설정해야 합니다.
At the bottom of the Google Cloud console, a
Cloud Shell
session starts and displays a command-line prompt. Cloud Shell is a shell environment
with the Google Cloud CLI
already installed and with values already set for
your current project. It can take a few seconds for the session to initialize.
[[["이해하기 쉬움","easyToUnderstand","thumb-up"],["문제가 해결됨","solvedMyProblem","thumb-up"],["기타","otherUp","thumb-up"]],[["이해하기 어려움","hardToUnderstand","thumb-down"],["잘못된 정보 또는 샘플 코드","incorrectInformationOrSampleCode","thumb-down"],["필요한 정보/샘플이 없음","missingTheInformationSamplesINeed","thumb-down"],["번역 문제","translationIssue","thumb-down"],["기타","otherDown","thumb-down"]],["최종 업데이트: 2025-09-04(UTC)"],[],[],null,["# Model Armor integration with Google Cloud services\n\n| **Preview**\n|\n|\n| This feature is subject to the \"Pre-GA Offerings Terms\" in the General Service Terms section\n| of the [Service Specific Terms](/terms/service-terms#1).\n|\n| Pre-GA features are available \"as is\" and might have limited support.\n|\n| For more information, see the\n| [launch stage descriptions](/products#product-launch-stages).\n\nModel Armor integrates with various Google Cloud services:\n\n- Google Kubernetes Engine (GKE) and Service Extensions\n- Vertex AI\n\nGKE and Service Extensions\n--------------------------\n\nModel Armor can be integrated with GKE through\nService Extensions. Service Extensions allow you to integrate\ninternal (Google Cloud services) or external (user-managed) services to process\ntraffic. You can configure a service extension on application load balancers,\nincluding GKE inference gateways, to screen traffic to and from a\nGKE cluster. This verifies that all interactions with the AI models\nare protected by Model Armor. For more information, see\n[Integration with GKE](/security-command-center/docs/model-armor-gke-integration).\n\nVertex AI\n---------\n\nModel Armor can be directly integrated into Vertex AI using either\n[floor settings](/security-command-center/docs/model-armor-vertex-integration#configure-floor-settings) or\n[templates](/security-command-center/docs/model-armor-vertex-integration#configure-templates).\nThis integration screens Gemini model requests and responses, blocking\nthose that violate floor settings. This integration provides prompt and response\nprotection within Gemini API in Vertex AI for the\n`generateContent` method. You need to enable Cloud Logging to get visibility\ninto the sanitization results of prompts and responses. For more information, see\n[Integration with Vertex AI](/security-command-center/docs/model-armor-vertex-integration).\n\nBefore you begin\n----------------\n\n### Enable APIs\n\nYou must enable Model Armor APIs before you can use Model Armor. \n\n### Console\n\n1.\n\n\n Enable the Model Armor API.\n\n\n [Enable the API](https://console.cloud.google.com/flows/enableapi?apiid=modelarmor.googleapis.com)\n\n \u003cbr /\u003e\n\n2. Select the project where you want to activate Model Armor.\n\n### gcloud\n\nBefore you begin, follow these steps using the Google Cloud CLI with the\nModel Armor API:\n\n1.\n\n\n In the Google Cloud console, activate Cloud Shell.\n\n [Activate Cloud Shell](https://console.cloud.google.com/?cloudshell=true)\n\n\n At the bottom of the Google Cloud console, a\n [Cloud Shell](/shell/docs/how-cloud-shell-works)\n session starts and displays a command-line prompt. Cloud Shell is a shell environment\n with the Google Cloud CLI\n already installed and with values already set for\n your current project. It can take a few seconds for the session to initialize.\n\n \u003cbr /\u003e\n\n2. Run the following command to set the API endpoint for the\n Model Armor service.\n\n ```bash\n gcloud config set api_endpoint_overrides/modelarmor \"https://modelarmor.\u003cvar translate=\"no\"\u003eLOCATION\u003c/var\u003e.rep.googleapis.com/\"\n ```\n\n Replace \u003cvar translate=\"no\"\u003eLOCATION\u003c/var\u003e with the region where you want to use Model Armor.\n\nRun the following command to enable Model Armor.\n\n\u003cbr /\u003e\n\n```bash\n gcloud services enable modelarmor.googleapis.com --project=PROJECT_ID\n \n```\n\n\u003cbr /\u003e\n\nReplace \u003cvar translate=\"no\"\u003ePROJECT_ID\u003c/var\u003e with the ID of the project.\n\nOptions when integrating Model Armor\n------------------------------------\n\nModel Armor offers the following integration options. Each option provides different\nfeatures and capabilities.\n\nFor the REST API integration option, Model Armor functions as a detector\nonly using templates. This means it primarily identifies and reports potential\npolicy violations based on predefined templates, rather than actively preventing\nthem.\n\nWith the Vertex AI integration option, Model Armor provides\ninline enforcement using floor settings or templates. This means\nModel Armor actively enforces policies by intervening directly\nin the process without requiring modifications to your application code.\n\nSimilar to Vertex AI, the GKE integration option also\noffers inline enforcement only using templates. This indicates that\nModel Armor can enforce policies directly within the inference gateway\nwithout requiring modifications to your application code."]]
