AI Verification Tech MapSearch

Mechanism · On-chip telemetry from timing, memory and performance counters

Sources

On this page
  1. BS. K. Monfared et al. (2026). Timing and Memory Telemetry on GPUs for AI Governance. arXiv. Source recordSupports: four timing and memory primitives, threat model, T4/H100 results, residency-test conditions, overheads, limitations · §3-§6, Figs. 5, 8, 10, 12, 15
  2. BJ. Petrie (2025). Guaranteeable Memory: An HBM-Based Chiplet for Verifiable AI Workloads. ICML 2025 Workshop on Technical AI Governance. Source recordSupports: guarantee chiplet under HBM observing memory traffic; HBM-standard compatibility; independence from the accelerator die · Abstract (read via ICML 2025 virtual site; OpenReview PDF not reachable)
  3. BR. Rahman & S. Tajdari (2026). Detecting Hidden ML Training With Zero-Overhead Telemetry. ICML 2026 Workshop on Technical AI Governance Research. Source recordSupports: NVML counter classifier, trust assumptions, GPU models, accuracy and evasion results, code statement, limitations · Abstract; threat model; results; limitations
  4. BG. Kulp et al. (2024). Hardware-Enabled Governance Mechanisms: Developing Technical Solutions to Exempt Items Otherwise Classified Under Export Control Classification Numbers 3A090 and 4A090. RAND Corporation. Source recordSupports: existing on-device counters and their use for metering · p. 19
  5. BA. O'Gara et al. (2025). Hardware-Enabled Mechanisms for Verifying Responsible AI Development. arXiv. Source recordSupports: candidate metering targets · §2.2.2, §2.5.2, Table 1
  6. CGloria Z (2026). On TEEs for Privacy-Preserving Monitoring in AI Governance. MIRI Technical Governance Team. Source recordSupports: counters as side channel; memory-residency and random challenges; completeness of workload declarations
  7. BNVIDIA (2025). NVIDIA Secure AI with Blackwell and Hopper GPUs (White Paper). NVIDIA documentation. Source recordSupports: performance counters disabled in full CC mode, and NVIDIA's side-channel rationale · p. 18
  8. AZ. Yang et al. (2024). Accurate and Convenient Energy Measurements for GPUs: A Detailed Study of NVIDIA GPU's Built-In Power Sensor. SC24: International Conference for High Performance Computing, Networking, Storage and Analysis. Source recordSupports: nvidia-smi power readings (via NVML) sample only 25% of runtime on A100 and H100; error about ±5% versus NVIDIA's claimed ±5 W · Abstract; accuracy findings
  9. BC. Shrauder & G. Frederick (2026). Introducing NVIDIA Fleet Intelligence for Real-Time GPU Fleet Visibility and Optimization. NVIDIA Technical Blog. Source recordSupports: NVIDIA Fleet Intelligence: general availability, read-only open-source host agent, telemetry collected, signed attestation evidence (provider self-description) · blog post

Search

Full search page