内容来源：https://research.google/blog/toward-provably-private-learning-from-federated-data/

内容总结：

谷歌近日宣布推出新一代联邦学习系统，通过引入可信执行环境（TEE），在提升训练速度、模型精度和设备覆盖范围的同时，实现可外部验证的隐私保障。

联邦学习是谷歌于2017年提出的一种机器学习技术，能够在分散的私有数据上训练模型，已广泛应用于Gboard下一词预测、智能撰写、Google Messages回复建议以及Android智能文本选择等功能。谷歌表示，其联邦学习系统开发遵循四项核心隐私原则：数据最小化、数据匿名化、透明度与控制权、可验证性与可审计性。

据介绍，新系统基于TEE技术，使运行其中的逻辑可被远程验证，第三方能够确认实际执行的程序，同时具备机密性和完整性。在此框架下，只有指标和差分隐私模型权重对工作负载操作者可见，设备上传的加密训练数据仅在TEE内解密和处理，且上传后仅在限定时间内可处理。设备端可获知可能访问其所上传数据的全部服务器工作负载，相关访问策略发布至公共透明日志Rekor，供外部审计人员监督。

谷歌表示，Gboard已采用这一新系统，并已基于该系统推出英语和日语下一词预测模型，隐私保障更强，模型精度也有所提升。新系统将计算瓶颈从设备端转移至服务器端，通过多机并行计算显著缩短训练时间。此前联邦学习模型训练通常需要一到两个月，受限于设备可用性、端侧算力及多任务资源竞争，而新系统下训练时间目前仅受TEE资源可用性限制。

谷歌还指出，新系统不仅可验证地执行联邦学习训练任务，也可运行其他可用Python表达的任意工作负载，目前正在探索合成数据生成等应用，并研究将执行任意Python的数据处理TEE与专注于大语言模型推理等功能的TEE结合使用。谷歌表示，未来TEE硬件的发展以及对侧信道观测的持续研究，有望为动态加载的工作负载提供更强保护，并期待此类系统最终能够提供差分隐私算法及系统组件软件实现的完整正确性证明。

中文翻译：

2026年10月2日
Katharine Daly，软件工程师；Daniel Ramage，研究总监，Google Research

我们宣布推出一套全新的联邦学习系统，该系统在将计算转移至服务器的同时，提供可被外部验证的隐私保障，从而提升训练速度、模型精度和设备覆盖范围。

2017年，Google 提出了联邦学习（Federated Learning，FL）——一种在去中心化的私有数据上训练模型的机器学习技术。它已被用于支撑诸多日常实用功能，包括 Gboard 上的下一词预测和智能撰写、Google Messages 中的回复建议，以及 Android 中的智能文本选择。

我们的联邦学习系统开发遵循四项核心隐私原则：（1）数据最小化，（2）数据匿名化，（3）透明度与用户控制，（4）可验证性与可审计性。多年来在匿名化方面的研究积累，通过矩阵分解 DP-FTRL（MF-DP-FTRL）等算法以及分布式差分隐私与安全聚合的结合，为生产模型带来了强有力的差分隐私（DP）保障。2025年，我们提出了基于这四项原则的联邦学习演进定义：

联邦学习（FL）是一种机器学习范式，多个实体（客户端）在服务提供方的协调下协作解决机器学习问题。一套完整的联邦学习系统应当使客户端能够对其数据、被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保持完全控制。联邦学习系统应当为数据由联邦学习客户端管理的用户提供适当的透明度和控制权。

在《Toward provably private learning from federated data》一文中，我们宣布了下一代联邦学习系统，该系统利用可信执行环境（TEE）来提供完全可验证、可审计的数据匿名化保障。在 TEE 中运行的逻辑可被远程证明（第三方可以验证正在执行的逻辑），同时获得机密性（其内部状态不可被观察）和完整性（逻辑不可被篡改），但受限于当前世代 TEE 的局限性。我们基于 TEE 的新联邦学习系统建立在 TEE 在单机层面所提供的这些属性之上，构建出一套完全可验证的端到端联邦学习系统，实现了更强的隐私保障和更高的模型精度。Gboard 已采用这一新系统，并从中受益——计算时间相比此前的联邦学习系统大幅缩短。

我们基于 TEE 的联邦学习系统建立在早期机密联邦分析和可证明私有洞察相关工作中所开发的技术之上。

该系统协调四个核心运行概念：

关于基于 TEE 的联邦学习系统设计的更多细节，请参阅我们的白皮书《Toward provably private learning from federated data》。

在早期的联邦学习系统中，设备数据被上传用于即时聚合，但外部观察者无法验证数据从未被记录或查看。后来，安全聚合使得上传数据可以通过密码学手段得到保护，但与最先进的中心化差分隐私保障不兼容。我们基于 TEE 的新系统代表了我们持续努力彻底消除对服务器运营方信任需求的又一里程碑。

在我们基于 TEE 的新联邦学习系统中，只有指标和差分隐私模型权重对工作负载运营方可见。从设备收集的加密训练数据只能在运行访问策略中所描述的 Python 训练程序的 TEE 内解密和处理，且仅限在上传后的一段有限时间内。

参与我们基于 TEE 的新联邦学习系统的设备知晓可能访问其上传数据的完整服务器工作负载集合。代表这些潜在未来服务器工作负载的访问策略被发布到 Rekor——一个公开的透明度日志，外部审计人员能够查看这些日志，以追踪设备可能参与的完整服务器端工作负载集合。

我们联邦学习系统中使用的 KMS 和数据处理二进制文件可以从机密联邦计算 GitHub 仓库中发布的开源代码进行可复现构建。

在我们早期的联邦学习系统中，服务器上运行的逻辑既无法被设备验证，也无法被审计人员验证，因此我们需要被信任能够正确地向梯度之和添加随机噪声以实现差分隐私。

在我们基于 TEE 的新联邦学习系统中，发布到 Rekor 的访问策略直接描述了表达联邦学习训练逻辑的 Python 程序。为在保护专有模型架构和数据预处理逻辑的同时保持可审计性，我们的数据处理 TEE 支持在运行时将序列化信息侧载到 Python 程序中。只要所有与隐私相关的逻辑仍然硬编码在 Python 程序中，这种侧载功能就允许必须保持专有的逻辑在 TEE 中运行，同时仍提供强有力的可外部验证的隐私保障（关于侧信道观测的讨论见下文）。

Gboard 已部署这套基于 TEE 的联邦学习系统，推出了英语和日语下一词预测模型，具备更强的隐私保障和更高的精度。这些改进可归因于新系统设计的多个方面。

通过在运行服务器端训练工作负载之前收集所有设备上传数据，我们不再需要担心设备可用性的昼夜变化影响训练进度。在服务器端执行时，我们可以在程序内动态计算最优的设备参与计划，并利用它来调整其他差分隐私参数。

过去，训练这些联邦学习模型每个可能需要一到两个月，进度受限于设备可用性、端上计算能力，以及多个训练工作负载对同一组设备资源的竞争。借助基于 TEE 的新系统，瓶颈已转移至服务器端，跨多台机器的计算并行化使我们能够实现训练时间的显著加速，目前仅受限于 TEE 资源的可用性。

在我们基于 TEE 的新联邦学习系统中，客户端梯度的计算被转移至服务器，解除了早期系统中端上计算资源的限制。这为使用联邦学习技术训练越来越大的模型铺平了道路。将 TEE 与加速器集成在此类应用场景中将发挥重要作用。

我们上述描述的系统不仅能够以可验证的方式执行联邦学习训练工作负载，还能执行任何可以用 Python 表达的任意工作负载。我们正在试验在此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。另一个探索方向是：将执行任意 Python 的数据处理 TEE 与我们在 LLM 推理等功能方面专精的其他数据处理 TEE 相结合。

这项工作朝着严格证明服务器端处理保护个人隐私迈出了一步。有了能够检查我们在 Google 运行的确切代码的外部验证者，我们能够提供强有力的保证：数据在服务器端的处理与所述完全一致。我们预期未来的 TEE 硬件，连同在缓解侧信道观测方面的持续研究，将为动态加载的工作负载提供更深入的保护，以抵御恶意服务器端攻击。我们预计，像我们这样的系统有朝一日可能附带对差分隐私算法和系统组件的软件实现的完整正确性证明。

作者感谢为基础设施设计和实现做出贡献的合作者：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 和 Zachary Garrett。我们还感谢支持 Gboard 集成的密切合作伙伴：Haicheng Sun、Heng Su、Jianpeng Hou、Liyang Jiang、Noriyuki Takahashi、Wenzhi Mao、Xiaojuan Fang、Yanxiang Zhang、Yingjie Liu、Yuanbo Zhang 和 Yun Wang。本工作得到了 Corinna Cortes、Shumin Zhai 和 Yossi Matias 的支持。我们还要感谢 Maysam Moussalem 对本文写作的反馈。

英文来源：

October 2, 2026
Katharine Daly, Software Engineer, and Daniel Ramage, Research Director, Google Research
We announce a new Federated Learning system that provides externally verifiable privacy guarantees while shifting computation to the server to improve training speed, accuracy, and device coverage.
In 2017, Google introduced Federated Learning (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 and Smart Compose on Gboard, reply suggestions in Google Messages, and Smart Text Selection in Android.
Our FL systems development is guided by four essential privacy principles: (1) data minimization, (2) data anonymization, (3) transparency and control, and (4) verifiability and auditability. Years of research development on anonymization have led to strong differential privacy (DP) guarantees for production models through algorithms like matrix factorization DP-FTRL (MF-DP-FTRL) and distributed differential privacy coupled with Secure Aggregation. In 2025, we introduced an evolved definition of FL centered on these four principles:
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.
In “Toward provably private learning from federated data”, we announce the next generation of our FL system, which leverages Trusted Execution Environments (TEEs) to provide fully verifiable and auditable data anonymization guarantees. Logic that runs in TEEs is remotely attestable (third parties can verify the logic that is being executed), and it also gains confidentiality (its internal state cannot be observed) and integrity (the logic cannot be disrupted), subject to current-generation TEE limitations. Our new TEE-based FL system builds on these properties which TEEs offer at the level of a single machine to form a fully verifiable end-to-end FL system that achieves stronger privacy guarantees and improved accuracy. Gboard has already adopted the new system and is benefiting from substantially faster compute times than our previous FL system.
Our TEE-based FL system builds on techniques developed in our earlier work on confidential federated analytics and provably private insights.
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.
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 allowed uploads to be protected cryptographically, but was not compatible with state-of-the-art central DP guarantees. 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, 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 Github repository.
In our earlier FL systems, the logic running on the server could be verified neither by devices nor auditors, and thus we needed to be trusted to correctly add random noise to gradient sums to provide differential privacy.
In our new TEE-based FL system, the access policies that are published to Rekor directly describe the Python program that expresses the FL training logic. To protect proprietary model architectures and data preprocessing logic while preserving auditability, our data processing TEEs support sideloading serialized information into the Python program at runtime. As long as all privacy-relevant logic remains hardcoded in the Python program, this sideloading functionality allows logic that must remain proprietary to run in the TEE while still providing strong externally verifiable privacy guarantees (see below for a discussion on side-channel observations).
Gboard has deployed this TEE-based FL system to launch English and Japanese next word prediction models with stronger privacy guarantees and improved accuracy. These improvements can be attributed to several aspects of the new system design.
By collecting all device uploads before running the server-side training workload, we no longer have to worry about diurnal variations 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.
This work is a step toward rigorous proof that server side processing preserves individual privacy. With external verifiers able to inspect exactly what code we run at Google, we are able to offer strong assurances that data is processed server-side exactly as described. We expect future TEE hardware, along with ongoing research into mitigating side-channel observations, to offer deeper protections for dynamically loaded workloads against malicious server-side attacks. We anticipate that systems like ours may one day come with full proofs of correctness of the software implementations of the DP algorithms and system components.
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.