{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-10-23T05:00:44.499Z","headline":"Google 发布基于 TEE 的下一代联邦学习系统，Gboard 已部署","description":"Google 宣布下一代联邦学习系统，利用可信执行环境（TEE）提供完全可验证、可审计的数据匿名化保证，访问策略发布至公共透明日志 Rekor，二进制可从开源代码可复现构建。Gboard 已部署该系统，用于英语和日语下一词预测模型，训练时间从过去的每次 1-2 个月显著缩短，并带来更强的隐私保证和更高的准确率。","url":"https://www.aioga.com/news/zfrloexd1672w3tse0utp4jgb/","mainEntityOfPage":"https://www.aioga.com/news/zfrloexd1672w3tse0utp4jgb/","datePublished":"2026-10-07T16:00:04.712Z","dateModified":"2026-10-07T16:00:04.712Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://research.google/blog/toward-provably-private-learning-from-federated-data","https://aihot.news/items/zfrloexd1672w3tse0utp4jgb"],"canonicalUrl":"https://www.aioga.com/news/zfrloexd1672w3tse0utp4jgb/","directAnswer":{"@type":"Answer","text":"Google Research宣布基于可信执行环境（TEE）的下一代联邦学习系统，并称Gboard已将其部署于英语和日语下一词预测模型。来源称，该系统可让访问策略发布至公共透明日志，二进制可由开源代码复现构建；训练时间较过去每次1至2个月显著缩短，准确率也有所提高。","url":"https://www.aioga.com/news/zfrloexd1672w3tse0utp4jgb/","dateCreated":"2026-10-07T16:00:04.712Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"Google Research：Blog source article","url":"https://research.google/blog/toward-provably-private-learning-from-federated-data","datePublished":"2026-10-07T16:00:04.712Z","provider":{"@type":"Organization","name":"Google Research：Blog","url":"https://research.google/blog/toward-provably-private-learning-from-federated-data"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.news/items/zfrloexd1672w3tse0utp4jgb","datePublished":"2026-10-07T16:00:04.712Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.news/items/zfrloexd1672w3tse0utp4jgb"}}],"aggregationSource":"Google Research：Blog","originalPublisher":{"name":"Google Research：Blog","url":"https://research.google/blog/toward-provably-private-learning-from-federated-data"},"geoDeepAnswer":null,"article":{"id":"zfrloexd1672w3tse0utp4jgb","slug":"zfrloexd1672w3tse0utp4jgb","url":"https://www.aioga.com/news/zfrloexd1672w3tse0utp4jgb/","title":"Google 发布基于 TEE 的下一代联邦学习系统，Gboard 已部署","title_en":"","summary":"Google 宣布下一代联邦学习系统，利用可信执行环境（TEE）提供完全可验证、可审计的数据匿名化保证，访问策略发布至公共透明日志 Rekor，二进制可从开源代码可复现构建。Gboard 已部署该系统，用于英语和日语下一词预测模型，训练时间从过去的每次 1-2 个月显著缩短，并带来更强的隐私保证和更高的准确率。","source":"Google Research：Blog","sourceUrl":"https://research.google/blog/toward-provably-private-learning-from-federated-data","aiHotUrl":"https://aihot.news/items/zfrloexd1672w3tse0utp4jgb","publishedAt":"2026-10-07T16:00:04.712Z","category":"行业动态","score":72,"selected":true,"articleBody":["Katharine Daly, Software Engineer, and Daniel Ramage, Research Director, Google Research","In 2017, Google introduced Federated Learning：https://research.google/blog/federated-learning-collaborative-machine-learning-without-centralized-training-data/ (FL) a machine learning technique that trains models across decentralized, private data. It has been used to power everyday helpful features, including next-word prediction：https://arxiv.org/abs/1811.03604 and Smart Compose：https://support.google.com/gboard/answer/7068415 on Gboard, reply suggestions：https://support.google.com/messages/answer/9327902?hl=en in Google Messages, and Smart Text Selection：https://research.google/blog/predicting-text-selections-with-federated-learning/ in Android.","Federated learning (FL) is a machine learning setting where multiple entities (clients) collaborate in solving a machine learning problem, under the coordination of a service provider. A complete FL system should enable clients to maintain full control over their data, the set of workloads allowed to access their data, and the anonymization properties of those workloads. FL systems should provide appropriate transparency and control to the users whose data is managed by FL clients.","Our TEE-based FL system builds on techniques developed in our earlier work on confidential federated analytics：https://research.google/blog/discovering-new-words-with-confidential-federated-analytics/ and provably private insights：https://research.google/blog/toward-provably-private-insights-into-ai-use/.","The system coordinates four core operational concepts:","For more details on the TEE-based FL system design please see our whitepaper, Toward provably private learning from federated data：https://arxiv.org/abs/2609.31494.","In earlier FL systems, device data was uploaded for the purpose of immediate aggregation, but there was no way for external observers to verify that the data was never logged or inspected. Later, Secure Aggregation：https://arxiv.org/abs/1611.04482 allowed uploads to be protected cryptographically, but was not compatible with state-of-the-art central DP guarantees：https://arxiv.org/abs/2202.08312. Our new TEE-based system represents the next milestone in our ongoing effort to completely remove the need to trust the server operator.","In our new TEE-based FL system, only metrics and differentially private model weights are visible to workload operators. Encrypted training data collected from devices can only be decrypted and processed within TEEs running Python training programs represented in the access policies, and only for a limited amount of time after upload.","Devices participating in our new TEE-based FL system know the full set of server workloads that may access data they’re uploading. The access policies representing these potential future server workloads are published to Rekor：https://docs.sigstore.dev/logging/overview/, a public transparency log, and external auditors are able to observe these logs to track the full set of server-side workloads that devices could potentially be participating in.","The KMS and data processing binaries used in our FL system can be reproducibly built from open source code published in the Confidential Federated Compute：https://github.com/google-parfait/confidential-federated-compute Github repository.","By collecting all device uploads before running the server-side training workload, we no longer have to worry about diurnal variations：https://arxiv.org/abs/1912.04977 in device availability impacting training progress. At server-side execution time, we can dynamically calculate the optimal device participation schedule within the program, and can use it to tune other DP parameters.","In the past, training these FL models could take 1-2 months each, with progress limited by device availability, on-device compute, and competition across multiple training workloads for the same set of device resources. With the new TEE-based system, bottlenecks have been moved to the server, and computation parallelization across many machines allows us to achieve significant speedups in training time, currently only limited by TEE resource availability.","In our new TEE-based FL system, computation of client gradients is shifted to the server, lifting limitations related to on-device compute resources that were present in earlier systems. This paves the way for training increasingly larger models using FL techniques. Integrating TEEs with accelerators will play an important role in such use cases.","The system we have described above is capable of executing not just FL training workloads in a verifiable manner, but also arbitrary workloads that can be expressed using Python. We are experimenting with running other types of workloads on this infrastructure, such as synthetic data generation workloads. Another area of exploration: using the data processing TEEs that execute arbitrary Python in combination with our other data processing TEEs that specialize in functionality such as LLM inference.","The authors would like to thank the collaborators who contributed to the infrastructure design and implementation: Arun Ganesh, Brendan McMahan, Brett McLarnon, Chunxiang (Jake) Zheng, Emily Glanz, Maya Spivak, Michael Reneer, Nova Fallen, Stefan Dierauf, Suxin Guo, Timon Van Overveldt, Yu Xiao, Zachary Charles, and Zachary Garrett. We also thank close partners who supported the Gboard integration: Haicheng Sun, Heng Su, Jianpeng Hou, Liyang Jiang, Noriyuki Takahashi, Wenzhi Mao, Xiaojuan Fang, Yanxiang Zhang, Yingjie Liu, Yuanbo Zhang, and Yun Wang. This work was supported by Corinna Cortes, Shumin Zhai, and Yossi Matias. We additionally thank Maysam Moussalem for feedback on the writing of this post."],"articleImages":[],"mediaStatus":"none","articleBodyZh":["Katharine Daly，软件工程师，Daniel Ramage，研究总监，谷歌研究院","2017 年，谷歌推出了联邦学习（Federated Learning, FL）：https://research.google/blog/federated-learning-collaborative-machine-learning-without-centralized-training-data/，这是一种在去中心化的私人数据上训练模型的机器学习技术。它已被用于支持日常有用功能，包括 Gboard 上的下一词预测：https://arxiv.org/abs/1811.03604 以及智能写作：https://support.google.com/gboard/answer/7068415，Google 信息应用中的回复建议：https://support.google.com/messages/answer/9327902?hl=en，以及 Android 上的智能文本选择：https://research.google/blog/predicting-text-selections-with-federated-learning/。","联邦学习（FL）是一种机器学习环境，其中多个实体（客户端）在服务提供者的协调下共同解决机器学习问题。完整的 FL 系统应允许客户端完全控制其数据、允许访问其数据的工作负载集合以及这些工作负载的匿名化属性。FL 系统应为被 FL 客户端管理数据的用户提供适当的透明度和控制权。","我们的基于 TEE 的 FL 系统建立在我们先前关于机密联邦分析（Confidential Federated Analytics）：https://research.google/blog/discovering-new-words-with-confidential-federated-analytics/ 和可证明私密洞察（Provably Private Insights）：https://research.google/blog/toward-provably-private-insights-into-ai-use/ 间开发的技术之上。","该系统协调四个核心操作概念：","有关基于 TEE 的 FL 系统设计的更多详细信息，请参见我们的白皮书《迈向可证明的联邦数据私密学习》：https://arxiv.org/abs/2609.31494。","在早期的 FL 系统中，设备数据被上传以用于即时聚合，但外部观察者无法验证数据是否从未被记录或检查。后来，安全聚合（Secure Aggregation）：https://arxiv.org/abs/1611.04482 允许通过加密保护上传的数据，但与最先进的集中式差分隐私保证（Central DP Guarantees）：https://arxiv.org/abs/2202.08312 不兼容。我们新的基于 TEE 的系统代表了我们在完全消除对服务器操作员信任需求方面持续努力的新里程碑。","在我们新的基于TEE的FL系统中，工作负载操作员只能看到指标和差分隐私模型权重。从设备收集的加密训练数据只能在运行访问策略中表示的Python训练程序的TEE内解密和处理，并且仅在上传后有限时间内可用。","参与我们新的基于TEE的FL系统的设备知道可能访问它们上传的数据的服务器工作负载的完整集合。表示这些潜在未来服务器工作负载的访问策略会发布到Rekor：https://docs.sigstore.dev/logging/overview/，这是一个公共透明日志，外部审计人员可以查看这些日志以跟踪设备可能参与的服务器端工作负载的完整集合。","我们FL系统中使用的KMS和数据处理二进制文件可以从发布在Confidential Federated Compute：https://github.com/google-parfait/confidential-federated-compute Github仓库中的开源代码可重复构建。","通过在运行服务器端训练工作负载之前收集所有设备的上传数据，我们不再需要担心设备可用性的昼夜变化：https://arxiv.org/abs/1912.04977 会影响训练进度。在服务器端执行时，我们可以在程序中动态计算最佳设备参与计划，并可用于调整其他DP参数。","过去，训练这些FL模型每次可能需要1-2个月，进度受设备可用性、设备计算能力以及多个训练工作负载竞争同一设备资源的限制。在新的基于TEE的系统中，瓶颈已转移到服务器端，并且通过多台机器的计算并行化使我们能够显著加快训练时间，目前仅受TEE资源可用性限制。","在我们新的基于TEE的FL系统中，客户端梯度的计算被转移到服务器端，从而消除了早期系统中与设备计算资源相关的限制。这为使用FL技术训练越来越大的模型铺平了道路。在此类用例中，将TEE与加速器集成将发挥重要作用。","我们上面描述的系统不仅能够以可验证的方式执行 FL（联邦学习）训练工作负载，还能执行所有可以用 Python 表达的任意工作负载。我们正在尝试在该基础设施上运行其他类型的工作负载，例如合成数据生成工作负载。另一个探索的方向是：将能够执行任意 Python 的数据处理 TEE 与我们专注于特定功能（如 LLM 推理）的其他数据处理 TEE 结合使用。","作者要感谢为基础设施设计和实施做出贡献的合作者：Arun Ganesh、Brendan McMahan、Brett McLarnon、Chunshan (Jake) Cheng、Emily Glanz、Maya Spivak、Michael Reneer、Nova Fallen、Stefan Dierauf、Suxinuo、Timon Van Overveldt、Yu Xiao、Zachary Charles 和 Zachary Garrett。我们还要感谢支持 Gboard 集成的密切合作伙伴：孙海成、苏恒、侯建鹏、蒋立阳、高桥纪之、毛文智、方晓娟、张彦翔、刘英杰、张远波和王云。这项工作得到了 Corinna Cortes、Shumin Zhai 和 Yossi Matias 的支持。我们还感谢 Maysam Moussalem 对本文撰写的反馈。"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"Google Research宣布基于可信执行环境（TEE）的下一代联邦学习系统，并称Gboard已将其部署于英语和日语下一词预测模型。来源称，该系统可让访问策略发布至公共透明日志，二进制可由开源代码复现构建；训练时间较过去每次1至2个月显著缩短，准确率也有所提高。","background":"来源介绍，早期联邦学习缺乏让外部观察者验证数据是否被记录或检查的方式；安全聚合虽能提供加密保护，但与先进的中心化差分隐私保证不兼容。新系统在TEE内解密和处理训练数据，工作负载运营者只能看到指标和差分隐私模型权重。","viewpoint":"Aioga 判断：该方案把访问策略透明记录、训练程序可复现构建与TEE内处理结合起来，重点强化了联邦学习的可验证性和审计基础。来源同时表明，数据处理受限于访问策略、指定程序和有限时间窗口，具体保障仍需结合公开材料核验。","implications":"可能影响：TEE和访问策略限制可能为联邦学习的数据使用提供更清晰的检查路径，也可能降低对服务器运营者的单纯信任依赖。但这不代表所有隐私风险均被消除，仍需要关注工作负载范围、差分隐私机制及实际部署边界。","nextStep":"后续观察：应关注公共透明日志中的访问策略、可复现构建材料和白皮书是否支持外部核验，并继续观察Gboard英语和日语下一词预测模型的部署范围。现有来源已披露训练时间显著缩短和准确率提高，但未给出具体缩短幅度或提升幅度。","evidenceRefs":["title","summary","articleBody","source"],"status":"published","aiGenerated":true,"autoApproved":true,"generatedBy":"aioga-editorial:gpt-5.6-sol","reviewedBy":"aioga-editorial-review:gpt-5.6-sol","generatedAt":"2026-10-03T17:17:46.397Z","sourceHash":"dffdc9bd7fd90e7f","review":{"approved":true,"groundedness":94,"clarity":91,"duplicationRisk":18,"blockingIssues":[],"notes":["“Aioga 判断”明确标注为观点，未冒充来源事实。","“可验证性和审计基础”“降低对服务器运营者的单纯信任依赖”等属于基于来源的合理推断，且使用了审慎措辞。","可将“下一词预测模型”统一为“下一词预测功能或模型”，但不影响事实准确性。"]},"validation":{"passed":true,"mode":"ai-auto","revisions":1,"checks":["schema","length","source-attribution","editorial-labels","inference-boundary","low-source-overlap","no-html","independent-ai-review"]}},"tags":["行业动态","Google Research：Blog"],"translations":{"zh-CN":{"title":"Google 发布基于 TEE 的下一代联邦学习系统，Gboard 已部署","summary":"Google 宣布下一代联邦学习系统，利用可信执行环境（TEE）提供完全可验证、可审计的数据匿名化保证，访问策略发布至公共透明日志 Rekor，二进制可从开源代码可复现构建。Gboard 已部署该系统，用于英语和日语下一词预测模型，训练时间从过去的每次 1-2 个月显著缩短，并带来更强的隐私保证和更高的准确率。","category":"行业动态","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google 发布基于 TEE 的下一代联邦学习系统，Gboard 已部署 - Aioga AI资讯","description":"Google 宣布下一代联邦学习系统，利用可信执行环境（TEE）提供完全可验证、可审计的数据匿名化保证，访问策略发布至公共透明日志 Rekor，二进制可从开源代码可复现构建。Gboard 已部署该系统，用于英语和日语下一词预测模型，训练时间从过去的每次 1-2 个月显著缩短，并带来更强的隐私保证和更高的准确率。","url":"https://www.aioga.com/news/zfrloexd1672w3tse0utp4jgb/","articleBody":["Katharine Daly，软件工程师，Daniel Ramage，研究总监，谷歌研究院","2017 年，谷歌推出了联邦学习（Federated Learning, FL）：https://research.google/blog/federated-learning-collaborative-machine-learning-without-centralized-training-data/，这是一种在去中心化的私人数据上训练模型的机器学习技术。它已被用于支持日常有用功能，包括 Gboard 上的下一词预测：https://arxiv.org/abs/1811.03604 以及智能写作：https://support.google.com/gboard/answer/7068415，Google 信息应用中的回复建议：https://support.google.com/messages/answer/9327902?hl=en，以及 Android 上的智能文本选择：https://research.google/blog/predicting-text-selections-with-federated-learning/。","联邦学习（FL）是一种机器学习环境，其中多个实体（客户端）在服务提供者的协调下共同解决机器学习问题。完整的 FL 系统应允许客户端完全控制其数据、允许访问其数据的工作负载集合以及这些工作负载的匿名化属性。FL 系统应为被 FL 客户端管理数据的用户提供适当的透明度和控制权。","我们的基于 TEE 的 FL 系统建立在我们先前关于机密联邦分析（Confidential Federated Analytics）：https://research.google/blog/discovering-new-words-with-confidential-federated-analytics/ 和可证明私密洞察（Provably Private Insights）：https://research.google/blog/toward-provably-private-insights-into-ai-use/ 间开发的技术之上。","该系统协调四个核心操作概念：","有关基于 TEE 的 FL 系统设计的更多详细信息，请参见我们的白皮书《迈向可证明的联邦数据私密学习》：https://arxiv.org/abs/2609.31494。","在早期的 FL 系统中，设备数据被上传以用于即时聚合，但外部观察者无法验证数据是否从未被记录或检查。后来，安全聚合（Secure Aggregation）：https://arxiv.org/abs/1611.04482 允许通过加密保护上传的数据，但与最先进的集中式差分隐私保证（Central DP Guarantees）：https://arxiv.org/abs/2202.08312 不兼容。我们新的基于 TEE 的系统代表了我们在完全消除对服务器操作员信任需求方面持续努力的新里程碑。","在我们新的基于TEE的FL系统中，工作负载操作员只能看到指标和差分隐私模型权重。从设备收集的加密训练数据只能在运行访问策略中表示的Python训练程序的TEE内解密和处理，并且仅在上传后有限时间内可用。","参与我们新的基于TEE的FL系统的设备知道可能访问它们上传的数据的服务器工作负载的完整集合。表示这些潜在未来服务器工作负载的访问策略会发布到Rekor：https://docs.sigstore.dev/logging/overview/，这是一个公共透明日志，外部审计人员可以查看这些日志以跟踪设备可能参与的服务器端工作负载的完整集合。","我们FL系统中使用的KMS和数据处理二进制文件可以从发布在Confidential Federated Compute：https://github.com/google-parfait/confidential-federated-compute Github仓库中的开源代码可重复构建。","通过在运行服务器端训练工作负载之前收集所有设备的上传数据，我们不再需要担心设备可用性的昼夜变化：https://arxiv.org/abs/1912.04977 会影响训练进度。在服务器端执行时，我们可以在程序中动态计算最佳设备参与计划，并可用于调整其他DP参数。","过去，训练这些FL模型每次可能需要1-2个月，进度受设备可用性、设备计算能力以及多个训练工作负载竞争同一设备资源的限制。在新的基于TEE的系统中，瓶颈已转移到服务器端，并且通过多台机器的计算并行化使我们能够显著加快训练时间，目前仅受TEE资源可用性限制。","在我们新的基于TEE的FL系统中，客户端梯度的计算被转移到服务器端，从而消除了早期系统中与设备计算资源相关的限制。这为使用FL技术训练越来越大的模型铺平了道路。在此类用例中，将TEE与加速器集成将发挥重要作用。","我们上面描述的系统不仅能够以可验证的方式执行 FL（联邦学习）训练工作负载，还能执行所有可以用 Python 表达的任意工作负载。我们正在尝试在该基础设施上运行其他类型的工作负载，例如合成数据生成工作负载。另一个探索的方向是：将能够执行任意 Python 的数据处理 TEE 与我们专注于特定功能（如 LLM 推理）的其他数据处理 TEE 结合使用。","作者要感谢为基础设施设计和实施做出贡献的合作者：Arun Ganesh、Brendan McMahan、Brett McLarnon、Chunshan (Jake) Cheng、Emily Glanz、Maya Spivak、Michael Reneer、Nova Fallen、Stefan Dierauf、Suxinuo、Timon Van Overveldt、Yu Xiao、Zachary Charles 和 Zachary Garrett。我们还要感谢支持 Gboard 集成的密切合作伙伴：孙海成、苏恒、侯建鹏、蒋立阳、高桥纪之、毛文智、方晓娟、张彦翔、刘英杰、张远波和王云。这项工作得到了 Corinna Cortes、Shumin Zhai 和 Yossi Matias 的支持。我们还感谢 Maysam Moussalem 对本文撰写的反馈。"]},"en":{"title":"Google releases a next-generation federated learning system based on TEE, now deployed in Gboard","summary":"Google announced its next-generation federated learning system, which uses trusted execution environments (TEEs) to provide fully verifiable and auditable guarantees for data anonymization. Access policies are published to the public transparency log Rekor, and the binaries can be reproducibly built from open-source code. Gboard has deployed this system for next-word prediction models in English and Japanese, significantly shortening training time from the previous 1–2 months per run while providing stronger privacy guarantees and higher accuracy.","category":"Industry","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google releases a next-generation federated learning system based on TEE, now deployed in Gboard - Aioga AI News","description":"Google announced its next-generation federated learning system, which uses trusted execution environments (TEEs) to provide fully verifiable and auditable guarantees for data anony...","url":"https://www.aioga.com/en/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:27:12.142Z"},"ja":{"title":"Google が TEE ベースの次世代フェデレーテッド ラーニング システムをリリース、Gboard がそれを導入","summary":"Google は、信頼できる実行環境 (TEE) を活用して、完全に検証可能で監査可能なデータ匿名化保証を提供する次世代のフェデレーテッド ラーニング システムを発表しました。アクセス ポリシーは公開の透過ログ Rekor に公開され、オープン ソース コードからバイナリを再現可能に構築できます。 Gboard は英語と日本語の次単語予測モデルにシステムを導入し、毎回のトレーニング時間を過去 1 ～ 2 か月から大幅に短縮し、その結果、プライバシーの保証が強化され、精度が向上しました。","category":"業界動向","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google が TEE ベースの次世代フェデレーテッド ラーニング システムをリリース、Gboard がそれを導入 - Aioga AIニュース","description":"Google は、信頼できる実行環境 (TEE) を活用して、完全に検証可能で監査可能なデータ匿名化保証を提供する次世代のフェデレーテッド ラーニング システムを発表しました。アクセス ポリシーは公開の透過ログ Rekor に公開され、オープン ソース コードからバイナリを再現可能に構築できます。 Gboard は英語と日本語の次単語予測モデルにシステムを導...","url":"https://www.aioga.com/ja/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:27:56.385Z"},"ko":{"title":"Google, TEE 기반 차세대 연합 학습 시스템 출시, Gboard 배포","summary":"Google은 TEE(Trusted Execution Environment)를 활용하여 완벽하게 검증 가능하고 감사 가능한 데이터 익명화를 보장하는 차세대 연합 학습 시스템을 발표했습니다. 액세스 정책은 공개 투명 로그 Rekor에 게시되며 바이너리는 오픈 소스 코드에서 재현 가능하게 구축될 수 있습니다. Gboard는 영어 및 일본어 다음 단어 예측 모델을 위한 시스템을 배포하여 매번 지난 1~2개월에 비해 훈련 시간을 크게 단축하고 더 강력한 개인정보 보호와 더 높은 정확도를 제공합니다.","category":"업계 동향","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google, TEE 기반 차세대 연합 학습 시스템 출시, Gboard 배포 - Aioga AI 뉴스","description":"Google은 TEE(Trusted Execution Environment)를 활용하여 완벽하게 검증 가능하고 감사 가능한 데이터 익명화를 보장하는 차세대 연합 학습 시스템을 발표했습니다. 액세스 정책은 공개 투명 로그 Rekor에 게시되며 바이너리는 오픈 소스 코드에서 재현 가능하게 구축될 수 있습니다. Gboard는...","url":"https://www.aioga.com/ko/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:27:53.734Z"},"es":{"title":"Google lanza el sistema de aprendizaje federado de próxima generación basado en TEE, Gboard ya está implementado","summary":"Google anunció el sistema de aprendizaje federado de próxima generación, que utiliza un Entorno de Ejecución Confiable (TEE) para proporcionar garantías de anonimización de datos completamente verificables y auditables, las políticas de acceso se publican en el registro público y transparente Rekor, y los binarios se pueden reconstruir a partir del código abierto. Gboard ha desplegado este sistema para modelos de predicción de la siguiente palabra en inglés y japonés, acortando significativamente el tiempo de entrenamiento que previamente era de 1-2 meses por cada vez, y ofreciendo mayores garantías de privacidad y una mayor precisión.","category":"Industria","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google lanza el sistema de aprendizaje federado de próxima generación basado en TEE, Gboard ya está implementado - Aioga Noticias de IA","description":"Google anunció el sistema de aprendizaje federado de próxima generación, que utiliza un Entorno de Ejecución Confiable (TEE) para proporcionar garantías de anonimización de datos c...","url":"https://www.aioga.com/es/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:33:42.090Z"},"fr":{"title":"Google publie un système de fédération d'apprentissage de nouvelle génération basé sur TEE, Gboard a déjà été déployé","summary":"Google a annoncé la prochaine génération de système d'apprentissage fédéré, utilisant un environnement d'exécution de confiance (TEE) pour fournir des garanties d'anonymisation des données entièrement vérifiables et auditables, avec les politiques d'accès publiées dans le registre public et transparent Rekor, et les binaires pouvant être reconstruits à partir du code source ouvert. Gboard a déployé ce système pour les modèles de prédiction du mot suivant en anglais et en japonais, réduisant considérablement le temps de formation de 1-2 mois par itération à une durée beaucoup plus courte et offrant de meilleures garanties de confidentialité ainsi qu'une plus grande précision.","category":"Industrie","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google publie un système de fédération d'apprentissage de nouvelle génération basé sur TEE, Gboard a déjà été déployé - Aioga Actualités IA","description":"Google a annoncé la prochaine génération de système d'apprentissage fédéré, utilisant un environnement d'exécution de confiance (TEE) pour fournir des garanties d'anonymisation des...","url":"https://www.aioga.com/fr/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:33:40.319Z"},"de":{"title":"Google veröffentlicht die nächste Generation des föderierten Lernsystems basierend auf TEE, Gboard wurde bereits implementiert","summary":"Google kündigte ein System für föderiertes Lernen der nächsten Generation an, das vertrauenswürdige Ausführungsumgebungen (TEE) nutzt, um vollständig überprüfbare und auditierbare Anonymisierungsgarantien für Daten bereitzustellen. Die Zugriffsrichtlinien werden im öffentlichen transparenten Protokoll Rekor veröffentlicht, und Binärdateien können aus dem Open-Source-Code reproduzierbar erstellt werden. Gboard hat dieses System für Vorhersagemodelle der nächsten Wörter in Englisch und Japanisch implementiert, wodurch die Trainingszeit im Vergleich zu früheren 1-2 Monaten pro Lauf deutlich verkürzt wurde und stärkere Datenschutzgarantien sowie höhere Genauigkeit erreicht werden.","category":"行业动态","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google veröffentlicht die nächste Generation des föderierten Lernsystems basierend auf TEE, Gboard wurde bereits implementiert - Aioga KI-News","description":"Google kündigte ein System für föderiertes Lernen der nächsten Generation an, das vertrauenswürdige Ausführungsumgebungen (TEE) nutzt, um vollständig überprüfbare und auditierbare...","url":"https://www.aioga.com/de/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:33:49.738Z"},"pt-BR":{"title":"Google lança sistema de aprendizado federado de próxima geração baseado em TEE, Gboard já implantado","summary":"O Google anunciou o próximo sistema de aprendizado federado, utilizando ambientes de execução confiáveis (TEE) para fornecer garantias de anonimização de dados totalmente verificáveis e auditáveis, com políticas de acesso publicadas no registro transparente público Rekor, e os binários podem ser reconstruídos a partir do código-fonte aberto. O Gboard já implementou esse sistema para modelos de previsão da próxima palavra em inglês e japonês, reduzindo significativamente o tempo de treinamento de 1-2 meses no passado, além de proporcionar maior garantia de privacidade e maior precisão.","category":"行业动态","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google lança sistema de aprendizado federado de próxima geração baseado em TEE, Gboard já implantado - Aioga Notícias de IA","description":"O Google anunciou o próximo sistema de aprendizado federado, utilizando ambientes de execução confiáveis (TEE) para fornecer garantias de anonimização de dados totalmente verificáv...","url":"https://www.aioga.com/pt-BR/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:39:47.940Z"},"ru":{"title":"Google представила систему федеративного обучения следующего поколения на базе TEE; Gboard уже развернут.","summary":"Google объявила о системе федеративного обучения нового поколения. Она использует доверенную среду выполнения (TEE), обеспечивая полностью проверяемую и подлежащую аудиту анонимизацию данных. Политики доступа публикуются в общедоступном журнале прозрачности Rekor, а двоичные файлы можно воспроизводимо собрать из открытого исходного кода. Gboard уже внедрила эту систему для моделей предсказания следующего слова на английском и японском языках. Время обучения, которое раньше составляло 1–2 месяца за цикл, значительно сократилось; при этом повысились гарантии конфиденциальности и точность.","category":"行业动态","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google представила систему федеративного обучения следующего поколения на базе TEE; Gboard уже развернут. - Aioga Новости ИИ","description":"Google объявила о системе федеративного обучения нового поколения. Она использует доверенную среду выполнения (TEE), обеспечивая полностью проверяемую и подлежащую аудиту анонимиза...","url":"https://www.aioga.com/ru/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:39:23.280Z"},"ar":{"title":"أصدرت جوجل نظام التعلم الفيدرالي من الجيل القادم القائم على TEE، وتم نشر Gboard","summary":"أعلنت Google عن نظام التعلم الفدرالي من الجيل التالي، يستخدم بيئة تنفيذ موثوقة (TEE) لتقديم ضمانات إخفاء الهوية للبيانات يمكن التحقق منها بالكامل ومراجعتها، ويتم نشر سياسات الوصول في سجل شفاف عام Rekor، ويمكن إعادة بناء الثنائيات من الكود المصدر المفتوح. تم نشر هذا النظام في Gboard لنموذج التنبؤ بالكلمة التالية باللغة الإنجليزية واليابانية، حيث تم تقصير وقت التدريب بشكل كبير من 1-2 شهر في كل مرة إلى وقت أقل، مع تقديم ضمانات خصوصية أقوى ودقة أعلى.","category":"行业动态","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"أصدرت جوجل نظام التعلم الفيدرالي من الجيل القادم القائم على TEE، وتم نشر Gboard - Aioga أخبار الذكاء الاصطناعي","description":"أعلنت Google عن نظام التعلم الفدرالي من الجيل التالي، يستخدم بيئة تنفيذ موثوقة (TEE) لتقديم ضمانات إخفاء الهوية للبيانات يمكن التحقق منها بالكامل ومراجعتها، ويتم نشر سياسات الوصول...","url":"https://www.aioga.com/ar/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:39:30.859Z"},"hi":{"title":"Google ने TEE आधारित अगले पीढ़ी के फेडरेटेड लर्निंग सिस्टम को जारी किया, Gboard पहले ही तैनात है","summary":"Google ने अगली पीढ़ी की संघीय शिक्षा प्रणाली की घोषणा की, जो भरोसेमंद निष्पादन वातावरण (TEE) का उपयोग करके पूरी तरह से सत्यापनीय और ऑडिट योग्य डेटा गुमनामी गारंटी प्रदान करती है, और एक्सेस नीतियों को सार्वजनिक पारदर्शी लॉग Rekor में प्रकाशित किया गया है, बाइनरी को ओपन-सोर्स कोड से पुनर्निर्मित किया जा सकता है। Gboard ने इस प्रणाली को अंग्रेजी और जापानी अगले शब्द की भविष्यवाणी मॉडल के लिए तैनात किया है, प्रशिक्षण समय पहले हर बार 1-2 महीने से उल्लेखनीय रूप से कम हो गया है, और यह अधिक मजबूत गोपनीयता गारंटी और उच्च सटीकता भी प्रदान करता है।","category":"行业动态","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google ने TEE आधारित अगले पीढ़ी के फेडरेटेड लर्निंग सिस्टम को जारी किया, Gboard पहले ही तैनात है - Aioga AI समाचार","description":"Google ने अगली पीढ़ी की संघीय शिक्षा प्रणाली की घोषणा की, जो भरोसेमंद निष्पादन वातावरण (TEE) का उपयोग करके पूरी तरह से सत्यापनीय और ऑडिट योग्य डेटा गुमनामी गारंटी प्रदान करती है, औ...","url":"https://www.aioga.com/hi/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:45:54.874Z"},"it":{"title":"Google ha rilasciato il sistema di apprendimento federato di nuova generazione basato su TEE, Gboard è già distribuito","summary":"Google ha annunciato il sistema di apprendimento federato di nuova generazione, che utilizza un ambiente di esecuzione attendibile (TEE) per fornire garanzie di anonimizzazione dei dati completamente verificabili e verificabili tramite audit, con le politiche di accesso pubblicate nel registro trasparente pubblico Rekor, e i binari ricostruibili dal codice open source. Gboard ha già implementato questo sistema per il modello di previsione della parola successiva in inglese e giapponese, riducendo significativamente il tempo di addestramento da 1-2 mesi per volta a tempi più brevi, offrendo al contempo maggiori garanzie di privacy e una precisione più elevata.","category":"行业动态","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google ha rilasciato il sistema di apprendimento federato di nuova generazione basato su TEE, Gboard è già distribuito - Aioga Notizie IA","description":"Google ha annunciato il sistema di apprendimento federato di nuova generazione, che utilizza un ambiente di esecuzione attendibile (TEE) per fornire garanzie di anonimizzazione dei...","url":"https://www.aioga.com/it/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:45:47.328Z"},"nl":{"title":"Google lanceert volgende generatie federatieve leersystemen gebaseerd op TEE, Gboard is al geïmplementeerd","summary":"Google kondigt het volgende generatie federale leersysteem aan, dat gebruikmaakt van Trusted Execution Environments (TEE) om volledig verifieerbare en auditeerbare gegevensanonimiseringsgaranties te bieden. Toegangsbeleid wordt gepubliceerd in het publieke transparantielogboek Rekor, en de binaries kunnen volledig uit de open source-code worden gereproduceerd. Gboard heeft dit systeem ingezet voor de voorspelling van het volgende woord in Engels en Japans, waardoor de trainingstijd van eerder 1-2 maanden per keer aanzienlijk is verkort, en het biedt sterkere privacygaranties en hogere nauwkeurigheid.","category":"行业动态","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google lanceert volgende generatie federatieve leersystemen gebaseerd op TEE, Gboard is al geïmplementeerd - Aioga AI-nieuws","description":"Google kondigt het volgende generatie federale leersysteem aan, dat gebruikmaakt van Trusted Execution Environments (TEE) om volledig verifieerbare en auditeerbare gegevensanonimis...","url":"https://www.aioga.com/nl/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:45:47.323Z"},"tr":{"title":"Google, TEE tabanlı yeni nesil federatif öğrenme sistemini yayımladı; sistem Gboard’da kullanıma sunuldu.","summary":"Google, güvenilir yürütme ortamları (TEE) kullanarak tamamen doğrulanabilir ve denetlenebilir veri anonimleştirme güvenceleri sağlayan yeni nesil bir federated learning sistemi duyurdu. Erişim politikaları, herkese açık Rekor şeffaflık günlüğünde yayımlanıyor ve ikili dosyalar açık kaynak koddan yeniden üretilebilir biçimde oluşturulabiliyor. Gboard bu sistemi, İngilizce ve Japonca sonraki kelime tahmin modelleri için kullanıma aldı. Eğitim süresi, geçmişte her seferinde 1-2 ay sürerken önemli ölçüde kısaldı; sistem ayrıca daha güçlü gizlilik güvenceleri ve daha yüksek doğruluk sağlıyor.","category":"行业动态","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google, TEE tabanlı yeni nesil federatif öğrenme sistemini yayımladı; sistem Gboard’da kullanıma sunuldu. - Aioga AI Haberleri","description":"Google, güvenilir yürütme ortamları (TEE) kullanarak tamamen doğrulanabilir ve denetlenebilir veri anonimleştirme güvenceleri sağlayan yeni nesil bir federated learning sistemi duy...","url":"https://www.aioga.com/tr/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:51:43.475Z"},"vi":{"title":"Google ra mắt hệ thống học liên kết thế hệ mới dựa trên TEE, đã được triển khai trên Gboard","summary":"Google công bố hệ thống học liên kết thế hệ tiếp theo, sử dụng môi trường thực thi đáng tin cậy (TEE) để cung cấp bảo đảm hoàn toàn có thể xác minh và kiểm toán đối với việc ẩn danh dữ liệu; các chính sách truy cập được công bố lên nhật ký minh bạch công khai Rekor, và tệp nhị phân có thể được xây dựng có thể tái tạo từ mã nguồn mở. Gboard đã triển khai hệ thống này cho các mô hình dự đoán từ tiếp theo bằng tiếng Anh và tiếng Nhật, rút ngắn đáng kể thời gian huấn luyện từ mức 1–2 tháng cho mỗi lần trước đây, đồng thời mang lại bảo đảm quyền riêng tư mạnh mẽ hơn và độ chính xác cao hơn.","category":"行业动态","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google ra mắt hệ thống học liên kết thế hệ mới dựa trên TEE, đã được triển khai trên Gboard - Tin tức AI Aioga","description":"Google công bố hệ thống học liên kết thế hệ tiếp theo, sử dụng môi trường thực thi đáng tin cậy (TEE) để cung cấp bảo đảm hoàn toàn có thể xác minh và kiểm toán đối với việc ẩn dan...","url":"https://www.aioga.com/vi/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:51:49.935Z"},"id":{"title":"Google merilis sistem pembelajaran federasi generasi berikutnya berbasis TEE, Gboard telah diterapkan","summary":"Google mengumumkan sistem federated learning generasi berikutnya, menggunakan Trusted Execution Environment (TEE) untuk memberikan jaminan anonimisasi data yang sepenuhnya dapat diverifikasi dan diaudit, dengan kebijakan akses diterbitkan ke log transparan publik Rekor, dan biner dapat dibangun kembali dari kode sumber terbuka. Gboard telah menerapkan sistem ini untuk model prediksi kata berikutnya dalam bahasa Inggris dan Jepang, dengan waktu pelatihan yang secara signifikan lebih singkat dari sebelumnya yang memakan waktu 1-2 bulan setiap sesi, serta memberikan jaminan privasi yang lebih kuat dan akurasi yang lebih tinggi.","category":"行业动态","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google merilis sistem pembelajaran federasi generasi berikutnya berbasis TEE, Gboard telah diterapkan - Berita AI Aioga","description":"Google mengumumkan sistem federated learning generasi berikutnya, menggunakan Trusted Execution Environment (TEE) untuk memberikan jaminan anonimisasi data yang sepenuhnya dapat di...","url":"https://www.aioga.com/id/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:51:40.088Z"},"th":{"title":"Google เปิดตัวระบบการเรียนรู้แบบรวมศูนย์รุ่นต่อไปที่อิงกับ TEE, Gboard ได้รับการติดตั้งแล้ว","summary":"Google ประกาศระบบการเรียนรู้แบบสหพันธ์รุ่นถัดไป ซึ่งใช้สภาพแวดล้อมที่เชื่อถือได้ในการประมวลผล (TEE) เพื่อให้การไม่ระบุตัวข้อมูลสามารถตรวจสอบและตรวจสอบได้อย่างสมบูรณ์ นโยบายการเข้าถึงถูกเผยแพร่ไปยังบันทึกสาธารณะโปร่งใส Rekor และไบนารีสามารถสร้างซ้ำได้จากซอร์สโค้ดเปิด Gboard ได้นำระบบนี้ไปใช้สำหรับโมเดลการทำนายคำถัดไปในภาษาอังกฤษและภาษาญี่ปุ่น เวลาในการฝึกจากแต่เดิมเดือนละครั้ง-สองเดือนถูกลดลงอย่างมาก และยังมอบการรับรองความเป็นส่วนตัวที่แข็งแรงขึ้นและความแม่นยำที่สูงขึ้น","category":"行业动态","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google เปิดตัวระบบการเรียนรู้แบบรวมศูนย์รุ่นต่อไปที่อิงกับ TEE, Gboard ได้รับการติดตั้งแล้ว - ข่าว AI Aioga","description":"Google ประกาศระบบการเรียนรู้แบบสหพันธ์รุ่นถัดไป ซึ่งใช้สภาพแวดล้อมที่เชื่อถือได้ในการประมวลผล (TEE) เพื่อให้การไม่ระบุตัวข้อมูลสามารถตรวจสอบและตรวจสอบได้อย่างสมบูรณ์ นโยบายการเข้าถ...","url":"https://www.aioga.com/th/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:57:45.416Z"},"pl":{"title":"Google opublikowało nowej generacji system uczenia federacyjnego oparty na TEE, Gboard został wdrożony","summary":"Google ogłosiło nową generację systemu uczenia federacyjnego, wykorzystującego zaufane środowiska wykonawcze (TEE) do zapewnienia w pełni weryfikowalnej i audytowalnej anonimowości danych, przy czym polityki dostępu są publikowane w publicznym, przejrzystym dzienniku Rekor, a pliki binarne można odtworzyć z kodu open source. Gboard wdrożył ten system do modeli przewidywania następnego słowa w języku angielskim i japońskim, skracając czas treningu z dotychczasowych 1-2 miesięcy do znacznie krótszego, jednocześnie oferując silniejszą ochronę prywatności i wyższą dokładność.","category":"行业动态","source":"Google Research：Blog","aggregationSource":"Google Research：Blog","pageTitle":"Google opublikowało nowej generacji system uczenia federacyjnego oparty na TEE, Gboard został wdrożony - Aioga Wiadomości AI","description":"Google ogłosiło nową generację systemu uczenia federacyjnego, wykorzystującego zaufane środowiska wykonawcze (TEE) do zapewnienia w pełni weryfikowalnej i audytowalnej anonimowości...","url":"https://www.aioga.com/pl/news/zfrloexd1672w3tse0utp4jgb/","contentTranslated":true,"translationStatus":"translated","translationRetryAt":"","translationError":"","sourceHash":"dd7b4231a5fa40bd","translatedAt":"2026-10-03T14:57:59.178Z"}},"evidenceTier":"verified-news","reviewStatus":"editorial-selected","indexable":true,"editorialCover":"/page-visuals/topic-timeline.png"}}