Source · Tier B · PreprintSource · Detecting Hidden ML Training With Zero-Overhead Telemetry
Detecting Hidden ML Training With Zero-Overhead Telemetry
R. Rahman, S. Tajdari. 2026. ICML 2026 Workshop on Technical AI Governance Research.
| Original | https://arxiv.org/abs/2606.19262 |
|---|---|
| DOI | 10.48550/arXiv.2606.19262 |
| arXiv | 2606.19262 |
| Version | v1 (2026-06-17) read on 2026-09-25. The paper header reads "Proceedings of the Second Workshop on Technical AI Governance Research (TAIGR) at the 43rd International Conference on Machine Learning, Seoul". Affiliations: R. Rahman (Machine Intelligence Research Institute), S. Tajdari (University of Virginia). |
| Accessed | 2026-10-20 |
| Imported from | hodgkins-ai-verification-papers@c71e59ff0e8e |
| Note | Listed under "Power telemetry and side-channel attacks" in the Hodgkins bibliography (CC BY 4.0). |
Cited by
- On-chip telemetry from timing, memory and performance countersMechanism
- Workload classification from telemetry and side channelsMechanism
- Declared hardware is idle or shut downClaim
- This compute runs inference, not trainingClaim
- Inference and training workloadsConcept
- Machine Intelligence Research InstituteOrganization