TEE 远程证明(以 Phala 为例)TEE remote attestation (Phala example)
交集:AI 推理跑在第三方云上时,用户怎么相信「模型与代码确实按声明运行」。TEE 远程证明把运行时度量、签名与验证报告交给软件核验。
适用范围:这里以 Phala 的 CPU/GPU TEE 与 attestation 示例说明流程;它不保证模型本身正确,也不保证所有云都支持同样的证明格式。
未解决:不同 TEE 厂商(Intel TDX / NVIDIA CC / AMD SEV)证明链与信任根不统一;侧信道、供应链与运维方仍可能破坏假设;跨云验证缺少统一标准。Intersection: when AI inference runs on a third-party cloud, how does the user know the model and code ran as claimed? TEE remote attestation exposes runtime measurements, signatures and verifier reports to software.
Scope: this uses Phala's CPU/GPU TEE and attestation example to show the flow; it does not guarantee the model is correct, nor that every cloud supports the same attestation format.
Open: trust roots and attestation chains differ across Intel TDX / NVIDIA CC / AMD SEV; side channels, supply chain and operators can still break the assumption; cross-cloud verification lacks a common standard.
关键参数Key specs
- 协议状态 / status
- 已部署(产品化示例)
- 证明对象 / attested
- 运行时度量 + 镜像摘要
- 验证方式 / verify
- 签名与 attestation 报告
- 来源 / source
- Phala 官方页
官方原文怎么说What the source says
TEE is a secure area inside a processor that protects code and data from the operating system, hypervisor, and other applications.
Instead of asking users to trust a cloud claim, Phala emits runtime measurements that software can verify.
runtime measurement
资料与许可Source and licence
- 资料时间Source checked
- 2026-11-19
- 许可 / 类型Licence / kind
- official-specification
- 许可说明Licence evidence
- https://phala.com/
- 类目Category
- 来源标识Provenance