P079 Independent verification of declared training-compute totals

Can any independent method - not relying on the chip's own secure enclave - verify a run's declared total compute on a real distributed run, to a stated error?

Independent verification of a run's declared FLOP total.

Status

Analyzed Proposals only, none run on a real distributed run: power-draw times energy-efficiency accounting of total operations (RAND 2025), workload-modelling of achievable utilisation (same source), spot-check zero-knowledge proofs with running cumulative-FLOP attestations (Peigné-Lefebvre, Nguyen & Wang 2026, 'a design, not a build'), and chip-logged weight snapshots with verifier retraining (Shavit 2023). No stated error for any of them on a real run; Ansari 2026 and Peigné-Lefebvre et al. both say no independent primitive for verifying reported training compute exists. Built but not on this quantity: Abdelghafar & Kulp 2026 classify training versus inference from real single-machine telemetry and say their duration tier 'is not a proxy for statutory training-compute thresholds'; Harack's Lawfare example is a code-inspection demonstration, not compute totals. Confidential-computing certificate rows depend on the chip's own enclave, which the headline excludes, so they are related, not evidence.

To verify the number of total operations done in given AI data centers in some period (from ta to tb ), one approach could be to measure AI chips’ power draw (P) over time (Watts per second), and then multiply it by the chips’ energy efficiency (OP per Watt)
Baker, Kulp, Marks, Brundage & Heim (2025), Verifying International Agreements on AI: Six Layers of Verification (RAND WR-A4077-1) (2025-07-25 (v2)), L3054–3057

Posed by

Also raised in: Petrie, Aarne, Ammann & Dalrymple (2025), Overview of Flexible Hardware-Enabled Guarantees (flexHEG Part 1); Petrie & Aarne (2025), Technical Options for Flexible Hardware-Enabled Guarantees (flexHEG Part 2); Abdelghafar & Kulp (2026), Privacy-Preserving AI Verification via Minimal Information Disclosure; Ansari (2026), Hardware-Level Governance of AI Compute: A Feasibility Taxonomy for Regulatory Compliance and Treaty Verification; Harack (2026), AI Verification: Infrastructure for Prosperity, Governance, and Peace (Lawfare); Wasil, Reed, Miller & Barnett (2024), Verification Methods for International AI Agreements; Peigné-Lefebvre, Nguyen & Wang (2026), Zero Knowledge Verification for Frontier AI Training Is Possible, arXiv 2606.05433.

Sub-questions

IDSub-questionWhat the sources report
P079.1Has any independent method (non-TEE, hardware- or facility-side, multi-device accounting) verified a declared run's FLOP total to a stated error on a real distributed run?partly answered
P079.2Can a TEE / confidential-computing attestation produce a 'training compute below a stated threshold' certificate checked by an independent verifier without exporting code, data or logs?partly answered
P079.3Can a low-leakage protocol let an inspector confirm 'no run above a stated FLOP threshold' (or RL vs non-RL) without seeing workload content, on multi-node runs?a source suggests an answer (not checked)

Possibly relevant work found by automated search

Found by AI agents searching for this problem's sub-questions, mostly by keyword and citation in the OpenAlex scholarly database and on the web; a second AI pass reviewed each one, and we spot-checked some by hand. They seemed reasonably related, as far as one can tell from outside each paper's field. Some come from neighbouring fields, such as arms control and financial audit, and bear on the problem by analogy. Not reviewed against the status above; a paper listed here does not change that status. 928 of 1052 quotes checked word for word against the paper's text.

For P079.1: Has any independent method (non-TEE, hardware- or facility-side, multi-device accounting) verified a declared run's FLOP total to a stated error on a real distributed run?

For P079.2: Can a TEE / confidential-computing attestation produce a 'training compute below a stated threshold' certificate checked by an independent verifier without exporting code, data or logs?

Data