Vertex AI 上的 Veo 在设计时考虑到了 Google 的 AI 原则。不过,开发者必须了解如何以负责任的方式安全测试和部署 Google 的模型。为了帮助开发者,Veo on Vertex AI 具有内置的安全功能,可帮助客户在其使用场景中阻止潜在的有害输出。如需了解详情,请参阅安全过滤条件。
我们建议客户在开发 AI 应用时使用公平性、可解释性、隐私权和安全最佳实践。如需了解详情,请参阅《人与 AI 指南》。
安全过滤器
Veo on Vertex AI 提供多种输入提示的方式来生成视频,包括文本、视频和图片。系统会根据一系列安全过滤条件评估提供给 Veo 的提示,这些过滤条件包括有害类别(例如 violence、sexual、derogatory 和 toxic)。这些安全过滤条件旨在过滤掉违反Google Cloud《平台使用政策》《生成式 AI 禁止的使用政策》或《我们的 AI 原则》的输入图片和视频。
如果模型响应请求并显示错误消息(例如 The
prompt couldn't be submitted 或 it might violate our
policies),则表示输入内容触发了安全过滤器。如果返回的视频少于请求的视频数,则由于不符合安全要求,某些生成的输出会被阻止。
安全过滤器代码类别
根据您配置的安全过滤器,您的输出可能包含类似于以下内容的安全代码:“Veo 无法生成视频,因为输入图片违反了 Vertex AI 的使用准则。如果您认为我们的处置结果有误,请发送反馈。支持代码:15236754”
[[["易于理解","easyToUnderstand","thumb-up"],["解决了我的问题","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["很难理解","hardToUnderstand","thumb-down"],["信息或示例代码不正确","incorrectInformationOrSampleCode","thumb-down"],["没有我需要的信息/示例","missingTheInformationSamplesINeed","thumb-down"],["翻译问题","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["最后更新时间 (UTC):2025-08-29。"],[],[],null,["# Responsible AI and usage guidelines for Veo\n\nVeo on Vertex AI is designed with Google's [AI\nprinciples](https://ai.google/responsibility/principles/) in mind. However, it\nis important for developers to understand how to test and deploy Google's models\nsafely and responsibly. To aid developers, Veo on Vertex AI has built-in safety\nfeatures to help customers block potentially harmful outputs within their use\ncases. For more information, see [safety filters](#safety-filters).\n\nWe encourage customers to use fairness, interpretability, privacy, and security\nbest practices when developing AI applications. For more information, see [The\nPeople + AI Guidebook](https://pair.withgoogle.com/guidebook/).\n\nSafety filters\n--------------\n\nVeo on Vertex AI offers several ways to input prompts to generate videos, including\ntext, video, and images. Prompts that are provided to Veo are\nassessed against a list of safety filters, which include *harmful categories*\n(for example, `violence`, `sexual`, `derogatory`, and `toxic`). These safety\nfilters aim to filter out input images and videos that violate the [Google Cloud\nPlatform Acceptable Use Policy](https://cloud.google.com/terms/aup?e=48754805),\n[Generative AI Prohibited Use\nPolicy](https://policies.google.com/terms/generative-ai/use-policy), or [Our AI\nPrinciples](https://ai.google/responsibility/principles/).\n\nIf the model responds to a request with an error message, such as `The\nprompt couldn't be submitted` or `it might violate our\npolicies`, then the input is triggering a safety filter. If fewer videos\nthan requested are returned, then some generated output is being blocked for not\nmeeting safety requirements.\n\nSafety filter code categories\n-----------------------------\n\nDepending on the safety filters that you configure, your output may contain a\nsafety code similar to: \"Veo could not generate videos because the input image\nviolates Vertex AI's usage guidelines. If you think this was an error, send\nfeedback. Support codes: **15236754**\"\n\nThe code listed in the output corresponds to a specific harmful category.\n\nThe following table displays the support code to safety category mappings:\n\nWhat's next\n-----------\n\n- Learn about [Responsible AI for Large Language Models\n (LLMs)](/vertex-ai/generative-ai/docs/learn/responsible-ai)\n\n- Learn more about Google's recommendations for [Responsible AI\n practices](https://ai.google/responsibilities/responsible-ai-practices/?category=general)\n\n- Read our blog, [A shared agenda for responsible AI\n progress](https://blog.google/technology/ai/a-shared-agenda-for-responsible-ai-progress/)"]]