大型生成式模型的出现给实现 Responsible AI 实践带来了新的挑战，因为它们可能具有开放式输出功能，并且有许多潜在的下游用途。除了 AI 原则之外，Google 还制定了《生成式 AI 使用限制政策》和面向开发者的生成式 AI 工具包。
Google 还就生成式 AI 模型提供了以下指南：
摘要
评估 AI 技术的公平性、问责性、安全性和隐私性，是负责任地构建 AI 的关键。这些检查应纳入产品生命周期的每个阶段，以确保为所有人开发安全、公平且可靠的产品。
下一项挑战
如未另行说明，那么本页面中的内容已根据知识共享署名 4.0 许可获得了许可，并且代码示例已根据 Apache 2.0 许可获得了许可。有关详情，请参阅 Google 开发者网站政策。Java 是 Oracle 和/或其关联公司的注册商标。
最后更新时间 (UTC)：2025-07-27。
[null,null,["最后更新时间 (UTC)：2025-07-27。"],[[["\u003cp\u003eGenerative AI models present new challenges to Responsible AI due to their open-ended output and varied uses, prompting the need for guidelines like Google's Generative AI Prohibited Use Policy and Toolkit for Developers.\u003c/p\u003e\n"],["\u003cp\u003eGoogle provides further resources on crucial aspects of generative AI, including safety, fairness, prompt engineering, and adversarial testing.\u003c/p\u003e\n"],["\u003cp\u003eBuilding AI responsibly requires thorough assessment of fairness, accountability, safety, and privacy throughout the entire product lifecycle.\u003c/p\u003e\n"],["\u003cp\u003eGoogle emphasizes the importance of Responsible AI and offers additional resources like the AI Principles, Generative AI information, and toolkits for developers.\u003c/p\u003e\n"]]],[],null,["# The next challenge\n\n\u003cbr /\u003e\n\nThe advent of large, generative models\nintroduces new challenges to implementing Responsible AI practices due to their\npotentially open-ended output capabilities and many potential downstream uses. In addition to the AI Principles, Google has a [Generative AI Prohibited Use Policy](https://policies.google.com/terms/generative-ai/use-policy)\nand [Generative AI Toolkit for Developers](https://ai.google.dev/responsible/docs).\n\nGoogle also offers guidance about generative AI models on:\n\n- [Safety](https://ai.google.dev/gemini-api/docs/safety-guidance)\n- [Prompt Engineering](/machine-learning/resources/prompt-eng)\n- [Adversarial Testing](/machine-learning/guides/adv-testing)\n\nSummary\n-------\n\nAssessing AI technologies for fairness, accountability, safety, and privacy is\nkey to building AI responsibly. These checks should be incorporated into every\nstage of the product lifecycle to ensure the development of safe, equitable, and\nreliable products for all.\n\nFurther learning\n----------------\n\n[Why we focus on AI -- Google AI](https://ai.google/why-ai/)\n\n[Google Generative AI](https://ai.google/discover/generativeai/)\n\n[PAIR Explorable: What Have Language Models Learned?](https://pair.withgoogle.com/explorables/fill-in-the-blank/)\n\n[Responsible AI Toolkit \\| TensorFlow](https://www.tensorflow.org/responsible_ai)"]]