aiqbench analysis
The paper proposes passive power metering as a way to verify whether large AI training runs are happening in a datacentre, without relying on the operator's cooperation. Its simulations suggest the method catches most training runs even when operators try to hide them, though the results come from simulated power data and a single adversarial model. For readers tracking AI governance, the notable point is that compute verification may be feasible with cheap, non-invasive instrumentation.
Detecting AI Training in Datacentre Power
arXiv:2610.30182v1 Announce Type: cross Abstract: AI governance agreements require methods to detect large model training runs on a monitored computing cluster, without trusting the operator. We develop a method to detect synchronous model training in facility-level power by Viterbi tracking of the wandering iteration cadence. On simulated datacentre power, when the operator makes no attempt to hide, the Viterbi tracker detects at least $96\%$ of training runs against a declared inference workload, at a $1\%$ false-alarm rate and within $0.04$ of the Neyman--Pearson optimum, given a meter that samples at $2$ Hz or faster. Of eight adversarial strategies an operator might use to hide, only extreme variation of the work per iteration evades the tracker. Its strongest form hides training in up to $73\%$ of five-minute observations, but over a day at least $87\%$ of such runs are still caught, at a $0.1\%$ false-alarm rate per run. This adversarial strategy costs no throughput, but in small-scale training the operator needs $93$--$267\%$ more compute to reach the same loss. A modest passive meter is sufficient to make hidden training both detectable and expensive, which makes power metering a promising part of compute verification that does not depend on trusting the operator.
Abstract:AI governance agreements require methods to detect large model training runs on a monitored computing cluster, without trusting the operator. We develop a method to detect synchronous model training in facility-level power by Viterbi tracking of the wandering iteration cadence. On simulated datacentre power, when the operator makes no attempt to hide, the Viterbi tracker detects at least $96\%$ of training runs against a declared inference workload, at a $1\%$ false-alarm rate and within $0.04$ of the Neyman--Pearson optimum, given a meter that samples at $2$ Hz or faster. Of eight adversarial strategies an operator might use to hide, only extreme variation of the work per iteration evades the tracker. Its strongest form hides training in up to $73\%$ of five-minute observations, but over a day at least $87\%$ of such runs are still caught, at a $0.1\%$ false-alarm rate per run. This adversarial strategy costs no throughput, but in small-scale training the operator needs $93$--$267\%$ more compute to reach the same loss. A modest passive meter is sufficient to make hidden training both detectable and expensive, which makes power metering a promising part of compute verification that does not depend on trusting the operator.
Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Machine Learning (cs.LG)
arXiv:2610.30182 [cs.CR]
(or arXiv:2610.30182v1 [cs.CR] for this version)
https://doi.org/10.48550/arXiv.2610.30182 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 298 words
