功耗能约束隐蔽算力吗?AI 治理中模拟验证的极限
Can Power Draw Constrain Covert Compute? Limits of Analogue Verification for AI Governance
AI 导读
一篇 arXiv 论文推导了封闭解 β,即功率曲线无法排除的最大隐藏计算量占申报机器容量的比例。基于 NVIDIA A100 GPU 的实测,最坏情况下 β=1.16,对抗性能量匹配调度至少隐藏 0.41,说明仅靠模拟功率测量对算力的约束很弱;在验证者可重执行申报工作的最强威胁模型下,β 可压到 0.059,为算力验证中模拟测量的作用给出了量化估计。
正文 · 原文
Abstract:Frontier AI treaties or agreements on limiting computation require external verification; an external auditor must be able to confirm how much computation actually ran and that parties are adhering to the agreement. Analogue, off-chip measurements such as power draw provide an information channel for verification. It is unknown how well these analogue channels can constrain the amount of computation against an adversary who actively tries to subvert the audit. We derive a closed form for $\beta$, the largest hidden computation a power trace cannot exclude, as a fraction of the declared machine capacity. Measurements on NVIDIA A100 GPUs give $\beta = 1.16$ in the worst case, while an adversarial matched-energy schedule hides at least $0.41$. Analogue power measurements alone therefore constrain compute weakly. Additional restrictions granted by the threat model, such as the ability of the verifier to re-execute the declared work at an observed operating point, let the verifier push $\beta$ down to $0.059$ in the maximally restricted case. This gives a quantitative estimate of what analogue measurements can contribute to compute verification.
| Comments: | 14 pages, 8 figures. v2: clarified the definition of $β$ and the relaxed bound; added related work on zero-knowledge inference proofs and proof-of-useful-work; results unchanged |
| Subjects: | Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR) |
| Cite as: | arXiv:2610.07476 [cs.CY] |
| (or arXiv:2610.07476v2 [cs.CY] for this version) | |
| https://doi.org/10.48550/arXiv.2610.07476 arXiv-issued DOI via DataCite |
Submission history
From: Tom Kimpson [view email]
[v1]
Mon, 5 Oct 2026 22:40:45 UTC (179 KB)
[v2]
Tue, 17 Nov 2026 06:14:15 UTC (180 KB)
来源:arXiv cs.AI · arxiv.org