Gemini Code Assist Standard 和 Enterprise 如何使用您的資料
透過集合功能整理內容
你可以依據偏好儲存及分類內容。
本文件說明 Gemini Code Assist Standard 版和 Enterprise 版在提供 AI 輔助功能時，如何履行 Google 對生成式 AI 技術所做的隱私權承諾。在開發環境中使用 Gemini Code Assist Standard 或 Enterprise 版本時，Google Cloud 會根據我們的服務條款和《Cloud 資料處理附加條款》處理提示。
如要進一步瞭解 Gemini Code Assist Standard 和 Enterprise 版，請參閱 Gemini Code Assist 總覽。
Google 的隱私權承諾
Google 是第一間發表 AI/機器學習隱私權承諾的公司，該文提到我們的信念：除了極致的安全性之外，客戶也應該對儲存在雲端的自家資料保有最大的掌控權。這項承諾也適用於 Gemini Code Assist Standard 和 Enterprise 版的生成式 AI 產品。Google 會透過完善的資料治理做法，確保團隊遵守這些承諾，包括審查 Google Cloud 在產品開發過程中使用的資料。如要進一步瞭解 Google 處理資料的方式，請參閱《客戶資料處理附加條款 (CDPA)》或適用於 Google Cloud 服務的資料處理協議。
您提交及接收的資料
您向 Gemini 提出的問題，包括您提交給 Gemini 以便分析或完成的任何輸入資訊或程式碼，都稱為提示。您從 Gemini 收到的答案或程式碼完成作業稱為「回覆」。
Gemini Code Assist Standard 版和 Enterprise 版不會使用提示或回覆內容做為模型的訓練資料。部分功能僅適用於 Gemini for Google Cloud「信任的測試人員」計畫，可讓您選擇性地分享資料，但這些資料會用於產品改善，而非訓練 Gemini 模型。
Gemini Code Assist Enterprise 中的程式碼自訂功能可讓您直接從 Gemini Code Assist 取得根據貴機構私人程式碼集產生的程式碼建議。您使用程式碼自訂功能時，我們會安全存取及儲存您的私人程式碼。這項存取權和儲存空間對於提供您要求的程式碼自訂服務至關重要。如要設定及使用程式碼自訂功能，請參閱「設定及使用 Gemini Code Assist 程式碼自訂功能」。
Gemini Code Assist 工具可讓開發人員不必離開 IDE，即可連結至外部服務，以便取得工作、摘要設計文件等。Gemini Code Assist 工具不會在工具之間共用資料。當您向某個工具傳送提示時，其他工具就無法存取該提示或回應。工具只能存取在提示中使用 @TOOL_NAME 語法直接傳送給工具的資料。
Gemini 仍在開發階段，因此可能會輸出看似合理卻與事實不符的內容。使用輸出內容前，請先確認內容是否屬實。詳情請參閱「Gemini Code Assist 和負責任的 AI 技術」。
提示訊息加密
向 Gemini 提交提示時，您的資料會在傳輸過程中加密，並輸入至 Gemini 中的底層模型。如要進一步瞭解 Gemini 資料加密功能，請參閱「預設靜態資料加密」和「傳輸中資料加密」。
由 Gemini 產生的節目資料
Gemini 會以第一方 Google Cloud 程式碼和所選第三方程式碼進行訓練。您必須負責程式碼的安全性、測試和效能，包括 Gemini 提供的任何程式碼完成、產生或分析功能。
如果建議內容直接從來源引用大量內容，Gemini 也會提供引用出處，協助您遵守授權規定。
由於 Gemini 中的回應是由經過多行程式碼訓練的模型產生，因此您應像處理其他程式碼一樣小心處理 Gemini 提供的程式碼。請務必妥善測試程式碼，並檢查是否有安全漏洞、不相容性和其他潛在問題。
後續步驟
除非另有註明，否則本頁面中的內容是採用創用 CC 姓名標示 4.0 授權，程式碼範例則為阿帕契 2.0 授權。詳情請參閱《Google Developers 網站政策》。Java 是 Oracle 和/或其關聯企業的註冊商標。
上次更新時間：2025-08-31 (世界標準時間)。
[null,null,["上次更新時間：2025-08-31 (世界標準時間)。"],[[["\u003cp\u003eGemini, an AI-powered assistant, adheres to Google's privacy commitment for generative AI technologies, ensuring high security and control over user data.\u003c/p\u003e\n"],["\u003cp\u003eGemini does not utilize user prompts or responses to train its models, maintaining the privacy of user interactions.\u003c/p\u003e\n"],["\u003cp\u003eCode customization features in Gemini securely access and store an organization's private code to provide tailored code suggestions.\u003c/p\u003e\n"],["\u003cp\u003eData submitted to Gemini is encrypted in-transit to ensure secure communication and protect sensitive information.\u003c/p\u003e\n"],["\u003cp\u003eUsers are responsible for the security, testing, and effectiveness of the code generated or suggested by Gemini, as it's trained on diverse code sources.\u003c/p\u003e\n"]]],[],null,["# How Gemini Code Assist Standard and Enterprise use your data\n\nThis document describes how Gemini Code Assist Standard and\nEnterprise editions, which offer AI-powered assistance, conform to\n[Google's privacy commitment](https://cloud.google.com/blog/products/ai-machine-learning/google-cloud-unveils-ai-and-ml-privacy-commitment)\nwith generative AI technologies. When you use Gemini Code Assist\nStandard or Enterprise editions in a development environment, Google Cloud\n[handles your prompts](#submit-receive-data) in accordance with our [terms of\nservice](https://cloud.google.com/terms) and [Cloud Data Processing Addendum](https://cloud.google.com/terms/data-processing-addendum).\n\nFor more information about Gemini Code Assist Standard and\nEnterprise editions, see the\n[Gemini Code Assist overview](/gemini-code-assist/docs/overview).\n\nGoogle's privacy commitment\n---------------------------\n\nGoogle was one of the first in the industry to publish an [AI/ML privacy\ncommitment](https://cloud.google.com/blog/products/ai-machine-learning/google-cloud-unveils-ai-and-ml-privacy-commitment),\nwhich outlines our belief that customers should have the highest level of\nsecurity and control over their data that's stored in the cloud. That commitment\nextends to Gemini Code Assist Standard and Enterprise edition\ngenerative AI products. Google helps ensure that its teams are following these\ncommitments through robust data governance practices, which include reviews of\nthe data that Google Cloud uses in the development of its products. You\ncan find more details about how Google processes data in\n[Customer Data Processing Addendum (CDPA)](https://cloud.google.com/terms/data-processing-addendum)\nor the data processing agreement applicable to your Google Cloud service.\n\nData you submit and receive\n---------------------------\n\nThe questions that you ask Gemini, including any input information or\ncode that you submit to Gemini to analyze or complete, are called\n*prompts* . The answers or code completions that you receive from Gemini\nare called *responses*.\n\nGemini Code Assist Standard and Enterprise editions don't use\nyour prompts or its responses as data to train its models. Some features are\nonly available through the\n[Gemini for Google Cloud Trusted Tester Program](https://cloud.google.com/gemini-for-cloud/ttp/welcome),\nwhich lets you optionally share data, but the data is used for product\nimprovements, not for training Gemini models.\n\n[Code customization](/gemini-code-assist/docs/code-customization-overview) in\nGemini Code Assist Enterprise lets you get code suggestions based\non your organization's private codebase directly from\nGemini Code Assist. When you use code customization, we securely\naccess and store your private code. This access and storage is essential for\ndelivering the code customization service you've requested. To configure and use\ncode customization, see\n[Configure and use Gemini Code Assist code customization](/gemini-code-assist/docs/code-customization).\n\n[Gemini Code Assist tools](/gemini-code-assist/docs/tools-agents/tools-overview)\nlet developers connect to external services without leaving the IDE in order to\nget tasks, summarize design documents and more. Gemini Code Assist\ntools don't share data between tools. When you send a prompt to one tool, other\ntools don't have access to that prompt or the response. Tools only have access\nto data sent directly to them using the `@TOOL_NAME` syntax in a prompt.\n\nBecause Gemini is an evolving technology, it can generate output that's\nplausible-sounding but factually incorrect. We recommend that you validate all\noutput from Gemini before you use it. For more information, see\n[Gemini Code Assist and responsible AI](/gemini-code-assist/docs/responsible-ai).\n\nEncryption of prompts\n---------------------\n\nWhen you submit prompts to Gemini, your data is encrypted in-transit as\ninput to the underlying model in Gemini. For more information on\nGemini data encryption, see\n[Default encryption at rest](https://cloud.google.com/docs/security/encryption/default-encryption)\nand [Encryption in transit](https://cloud.google.com/docs/security/encryption-in-transit).\n\nProgram data generated from Gemini\n----------------------------------\n\nGemini is trained on first-party Google Cloud code as well as\nselected third-party code. You're responsible for the security, testing, and\neffectiveness of your code, including any code completion, generation, or\nanalysis that Gemini offers you.\n\nGemini also provides source citations when suggestions directly quote\nat length from a source to help you comply with any license requirements.\n\nBecause responses in Gemini are generated from a model that's trained\non many lines of code, you should exercise the same care with\nGemini-provided code that you would with any other code. Make sure that\nyou test the code properly and check for security vulnerabilities,\nincompatibilities, and other potential issues.\n\nWhat's next\n-----------\n\n- Learn about the [security, privacy, and compliance of Gemini Code Assist](https://cloud.google.com/gemini/docs/codeassist/security-privacy-compliance)."]]
