# AI verification proposal

A proposal built with the Proposal Explorer of the AI Verification Tech Map (https://trustbutveri.fyi/), from its records of 2026-10-09. Interactive version: https://trustbutveri.fyi/explorer/?mechanisms=M-0003,M-0010,M-0012&hide=weights

How to read it: a claim is something one party wants to verify about another's AI hardware or software. A mechanism is a general technique for verifying claims; it is "aimed at" a claim when that is its direct purpose, and "supporting" when it contributes without being aimed at it. A claim is addressed when a mechanism in the proposal is aimed at it and is not excluded by the filters; addressed does not mean verified, so check that mechanism's development status, security evidence and findings. Definitions: https://trustbutveri.fyi/about/methodology/ (roles, properties and findings) and https://trustbutveri.fyi/about/readiness/ (development status).

## Filters

Filters apply to mechanisms only and describe the setting the proposal is for.

- **Keep hidden from the verifier: model weights.** Removes mechanisms that show the asset to the verifier. Conditional or unspecified exposure stays with a note and needs checking against the privacy requirement. Model weights: the checked model's parameters. Inputs and outputs: the requests a deployed model serves and its responses. Training data: what a model was trained on. Each mechanism's exposure is the editors' reading of its record: shown, depends on the design (kept, with a note), hidden, not involved, or unspecified for a selected implementation. Code and configuration are not covered yet.

24 of 25 mechanisms on the map pass these filters.

## Overview

One row per mechanism, read from its record. Open failures: critical / significant / minor. The last three columns are the editors' reading of what the verifier sees. Findings are grouped as known failures, scope limitations and open questions. Only known failures count as failures. Counts are an inventory of published findings, not a risk score.

| Mechanism | Development | Security evidence | Prover | Attack testing | Hardware | Open failures | Weights | Inputs and outputs | Training data |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Whole-workload recomputation (reproducible packets) | Proposed | No published adversarial analysis recorded | Adversarial | None | Retrofit device | 0 / 0 / 0 | depends | depends | depends |
| On-chip telemetry from timing, memory and performance counters | Research demonstration | Published attack testing | Semi-trusted | Red-teamed | Existing features | 0 / 2 / 0 | depends | depends | depends |
| Hardware-attested weight binding | Operational use | Published attack testing | Semi-trusted | Independent red-team | Existing features | 1 / 0 / 0 | hidden | unspecified | not involved |

## Claims

No claims chosen.

## Mechanisms

### Whole-workload recomputation (reproducible packets)

Organizing all AI workloads in a facility into discrete, reproducible units, so that a verifier can recompute a random sample and check each one. ([Whole-workload recomputation (reproducible packets)](https://trustbutveri.fyi/mechanisms/reproducible-computation-packets/))

- Assessment: mechanism family.
- Development: Proposed (legacy code R1), assessed for recomputing whole workloads to show a cluster runs only declared inference.
- Security evidence: No published adversarial analysis recorded. Independent evaluation: unassessed. Formal proof: unassessed. Deployment assurance: unassessed.
- Claims in this proposal: none of them.
- Threat model: adversarial prover. Hardware: retrofit device. Prover cooperation: required. Attack testing: none. Category: Isolation & system architectures.
- What the verifier sees: model weights depends; inputs and outputs depends; training data depends. Recomputing sampled units needs weights and sampled inputs or training data inside the checking environment. The design depends on securing that environment; disclosure depends on its confidentiality boundary.
- Filter note: May show model weights, depending on the design. Recomputing sampled units needs weights and sampled inputs or training data inside the checking environment. The design depends on securing that environment; disclosure depends on its confidentiality boundary.

### On-chip telemetry from timing, memory and performance counters

Uses on-chip measurements, such as task timings, whether data sits in chip memory, and performance counters, as evidence of what AI chips are running. ([On-chip telemetry from timing, memory and performance counters](https://trustbutveri.fyi/mechanisms/on-chip-telemetry/))

- Assessment: mechanism family.
- Development: Research demonstration (legacy code R2), assessed for workload evidence from GPU counters and timing, assuming authentic measurements.
- Security evidence: Published attack testing. Independent evaluation: unassessed. Formal proof: unassessed. Deployment assurance: unassessed.
- Claims in this proposal: none of them.
- Threat model: semi-trusted prover. Hardware: existing features. Prover cooperation: partial. Attack testing: red-teamed. Category: On-chip & hardware-enabled.
- What the verifier sees: model weights depends; inputs and outputs depends; training data depends. Counters do not read weights or data, but richer counters can leak secrets through side channels.
- Filter note: May show model weights, depending on the design. Counters do not read weights or data, but richer counters can leak secrets through side channels.

### Hardware-attested weight binding

Checks that a hardware enclave serves committed model weights, by attesting software that tests the weights against a hash commitment when they are read. ([Hardware-attested weight binding](https://trustbutveri.fyi/mechanisms/model-identity-attestation/))

- Assessment: mechanism family.
- Development: Operational use (legacy code R3), assessed for hardware-attested weight binding showing users that a service runs its committed weights.
- Security evidence: Published attack testing. Independent evaluation: unassessed. Formal proof: unassessed. Deployment assurance: unassessed.
- Claims in this proposal: none of them.
- Threat model: semi-trusted prover. Hardware: existing features. Prover cooperation: required. Attack testing: independent red-team. Category: Cryptographic & computational.
- What the verifier sees: model weights hidden; inputs and outputs unspecified; training data not involved. Verifiers check a hash commitment to the weights, carried in a hardware attestation, and need no access to the weights themselves. The mechanism does not specify whether prompts and outputs are disclosed to a verifier. It involves no training data.


## Properties

**Operational use**

- Hardware-attested weight binding: Operational use (legacy code R3), assessed for hardware-attested weight binding showing users that a service runs its committed weights

**Built for an adversarial prover**

- Whole-workload recomputation (reproducible packets)

**No new hardware needed**

- On-chip telemetry from timing, memory and performance counters
- Hardware-attested weight binding

**Failures since mitigated**

- Launch-state attestation does not by itself cover weights loaded later (in Hardware-attested weight binding) [10][12]


## Attack testing

Attack testing records published testing for this use. It does not by itself show independent review, a formal proof or that a deployed system is secure.

**Testing history**

- On-chip telemetry from timing, memory and performance counters: Red-teamed
- Hardware-attested weight binding: Independent red-team


## Limits

**Open critical failures**

- Underlying attestation can be forged or relayed (known failure, demonstrated attack, in Hardware-attested weight binding; https://trustbutveri.fyi/mechanisms/model-identity-attestation/evidence/flaws/1/) [13][14][18][19][20][21]. Inherited finding. Critical for weight binding against an operator with physical access to affected hardware, or with control of the hypervisor on an AMD SEV-SNP platform without AMD's fixes. PAL*M excludes physical attacks, and Tinfoil acknowledges this boundary. The enclave route inherits the platform-specific TEE attestation failures. Intel TDX forgery and H100 relay were demonstrated with physical access and host control. Battering RAM defeated AMD SEV-SNP attestation on DDR4 servers; RMPocalypse did so from malicious host software on platforms without AMD's fixes. These demonstrate failures of the trust roots, not of each model-commitment protocol. Related finding: https://trustbutveri.fyi/mechanisms/tee-remote-attestation/evidence/flaws/1/.

  Response: The TEE.fail authors report that physical interposer attacks are outside Intel's and AMD's threat models. AMD reports fixes for RMPocalypse.

  Related mechanism: Hardware-enabled guarantees (flexHEG) and guarantee processors (R1, not in the proposal). A tamper-protected enclosure around the chip is the proposed answer when the party that holds the hardware may attack it physically.

**Open significant failures**

- Adversarially disguised fine-tuning partly evades classification (known failure, demonstrated attack, in On-chip telemetry from timing, memory and performance counters; https://trustbutveri.fyi/mechanisms/on-chip-telemetry/evidence/flaws/2/) [9]. Across 20 evasion strategy families in five monitor-evader rounds, the classifier's accuracy against the most challenging disguised workloads held out from training was 43–87%. White-box LoRA fine-tuning was the only evasion family detected less than half the time. The evaluation covered single nodes, PyTorch-level evasion and NVIDIA hardware.
- Counters leak information about protected workloads (known failure, theoretical argument, in On-chip telemetry from timing, memory and performance counters; https://trustbutveri.fyi/mechanisms/on-chip-telemetry/evidence/flaws/4/) [10][11]. Performance counters have been used as a side channel against TEEs, for example in CounterSEVeillance. NVIDIA disables performance counters in full confidential-computing mode, stating that they could provide an avenue for side-channel attacks. Richer counters for verification therefore pull against confidentiality.

**Scope limitations**

- Spare compute is outside the scheme (scope limitation, theoretical argument, in Whole-workload recomputation (reproducible packets); https://trustbutveri.fyi/mechanisms/reproducible-computation-packets/evidence/flaws/1/) [1][2]. The plan states that it does not verify that spare compute is not used for unapproved workloads, because this seems very challenging. Recomputation checks the correctness of declared work, not its completeness.

  Related mechanism: Proofs of useful work for capacity accounting (R1, not in the proposal). Proposed as one input to accounting for spare capacity on declared hardware.
- Software-read telemetry can be forged by the operator (scope limitation, theoretical argument, in On-chip telemetry from timing, memory and performance counters; https://trustbutveri.fyi/mechanisms/on-chip-telemetry/evidence/flaws/1/) [7][9]. NVML-based classification assumes trustworthy telemetry. Without a tamper-resistant read path, an authenticated telemetry channel and secure boot of the monitoring software, an operator who controls the full software stack could forge counter values. Monfared et al. start from the same premise: current GPUs expose little trusted telemetry and can be modified or virtualized.

  Related mechanism: Hardware-enabled guarantees (flexHEG) and guarantee processors (R1, not in the proposal). A guarantee processor on the chip would give the tamper-resistant, authenticated telemetry path the flaw says is missing.
- Timing challenges do not identify the individual chip (scope limitation, theoretical argument, in On-chip telemetry from timing, memory and performance counters; https://trustbutveri.fyi/mechanisms/on-chip-telemetry/evidence/flaws/3/) [7]. GEMM and VDF challenges can be answered by identical GPUs elsewhere, and floating-point fingerprints distinguish GPU models, not individual devices. GPU virtualization adds timing leakage that prevents attributing compute use.
- For private models, a user can confirm consistency but not content (scope limitation, open question, in Hardware-attested weight binding; https://trustbutveri.fyi/mechanisms/model-identity-attestation/evidence/flaws/3/) [12][22]. When weights are not published, users can check that the same root hash is served each time, but not what the model is. Pairing the hash with an attested evaluation, as in Attestable Audits, is one proposed remedy.

**Open questions**

- Non-compliant work could be encoded inside compliant-looking packets (open question, theoretical argument, in Whole-workload recomputation (reproducible packets); https://trustbutveri.fyi/mechanisms/reproducible-computation-packets/evidence/flaws/2/) [1]. The plan notes that an AI company might try to encode a non-compliant workload inside a workload that looks compliant on the surface.
- No quantified error rates or formal thresholds for timing primitives (open question, open question, in On-chip telemetry from timing, memory and performance counters; https://trustbutveri.fyi/mechanisms/on-chip-telemetry/evidence/flaws/5/) [7]. Monfared et al. state that false-positive and false-negative rates are not quantified and leave hardware-specific formal thresholds to future work.

**Not yet demonstrated**

- Whole-workload recomputation (reproducible packets): Proposed (legacy code R1), assessed for recomputing whole workloads to show a cluster runs only declared inference


## Possible additions

Mechanisms on the map, not in the proposal, that the records connect to an unaddressed or partly addressed claim, an open failure or a dependency. Pointers, not recommendations: each brings its own readiness level and findings, and none is claimed to close a failure.

- **Hardware-enabled guarantees (flexHEG) and guarantee processors** (Proposed (legacy code R1), assessed for checking and enforcing training-compute limits on chips, against adversaries up to states)
  - Bears on the open critical failure "Underlying attestation can be forged or relayed" in Hardware-attested weight binding. A tamper-protected enclosure around the chip is the proposed answer when the party that holds the hardware may attack it physically.
  - On-chip telemetry from timing, memory and performance counters waits on it: Shipping accelerators need a tamper-resistant, authenticated telemetry path.
- **Deterministic and bit-exact inference** (Operational use (legacy code R3), assessed for reproducing open-model inference from receipts in Gensyn's information-market service)
  - Whole-workload recomputation (reproducible packets) waits on it: Workloads are not reproducible by default, and achieving reproducibility may cost performance.
- **TEE remote attestation for AI workloads** (Operational use (legacy code R3), assessed for showing which software ran to a party that distrusts the operator holding the hardware)
  - Hardware-attested weight binding waits on it: Attestation that resists physical attackers, for the enclave variant.
  - On-chip telemetry from timing, memory and performance counters depends on it.
- **Network taps and certifiers** (Proposed (legacy code R1), assessed for committing a complete record of cluster traffic, so declared inference can be checked)
  - Whole-workload recomputation (reproducible packets) waits on it: All traffic must reach the recomputation server via network taps, and the server's integrity is critical.


## Dependencies

**Missing prerequisites**

- Deterministic and bit-exact inference (Operational use (legacy code R3), assessed for reproducing open-model inference from receipts in Gensyn's information-market service), needed by Whole-workload recomputation (reproducible packets)
- Network taps and certifiers (Proposed (legacy code R1), assessed for committing a complete record of cluster traffic, so declared inference can be checked), needed by Whole-workload recomputation (reproducible packets)
- TEE remote attestation for AI workloads (Operational use (legacy code R3), assessed for showing which software ran to a party that distrusts the operator holding the hardware), needed by On-chip telemetry from timing, memory and performance counters and Hardware-attested weight binding

**Shared foundations**

- TEE remote attestation for AI workloads, relied on by On-chip telemetry from timing, memory and performance counters and Hardware-attested weight binding

**Blockers**

- Whole-workload recomputation (reproducible packets): Workloads are not reproducible by default, and achieving reproducibility may cost performance. (performance & compatibility; waits on Deterministic and bit-exact inference) [1]
- Whole-workload recomputation (reproducible packets): Network packets are not individually reproducible by default; making them so may need considerable software, firmware and hardware work. Amodo rates this 'not on track'. (performance & compatibility) [4]
- Whole-workload recomputation (reproducible packets): All traffic must reach the recomputation server via network taps, and the server's integrity is critical. (hardware trust; waits on Network taps and certifiers) [1][4]
- Whole-workload recomputation (reproducible packets): Recomputing training steps needs checkpoints: writing one at every step would cost more than 100% overhead, so Amodo's design needs a spare data-parallel replica that tracks the weights instead. (performance & compatibility) [2]
- On-chip telemetry from timing, memory and performance counters: Shipping accelerators need a tamper-resistant, authenticated telemetry path. (hardware trust; waits on Hardware-enabled guarantees (flexHEG) and guarantee processors) [8][9]
- On-chip telemetry from timing, memory and performance counters: NVIDIA's full confidential-computing mode disables the hardware performance counters its profiling tools use, so telemetry that needs them conflicts with it. (privacy & leakage) [10][11]
- On-chip telemetry from timing, memory and performance counters: Continuous challenge puzzles cost power and throughput on production workloads. (performance & compatibility) [7]
- On-chip telemetry from timing, memory and performance counters: Evaluation has not gone beyond single nodes, framework-level evasion and one vendor's hardware. (adversarial validation) [9]
- Hardware-attested weight binding: Attestation that resists physical attackers, for the enclave variant. (hardware trust; waits on TEE remote attestation for AI workloads) [18]


## What the verifier sees

- Model weights: shown by none; depends on the design for Whole-workload recomputation (reproducible packets) and On-chip telemetry from timing, memory and performance counters; hidden by Hardware-attested weight binding; not involved in none; unspecified for none.
- Inputs and outputs: shown by none; depends on the design for Whole-workload recomputation (reproducible packets) and On-chip telemetry from timing, memory and performance counters; hidden by none; not involved in none; unspecified for Hardware-attested weight binding.
- Training data: shown by none; depends on the design for Whole-workload recomputation (reproducible packets) and On-chip telemetry from timing, memory and performance counters; hidden by none; not involved in Hardware-attested weight binding; unspecified for none.

## Implementations

- Whole-workload recomputation (reproducible packets): [AI 2040 inference-only verification stack](https://trustbutveri.fyi/implementations/ai-2040-inference-only-verification-plan/) (R1, proposed architecture)
- On-chip telemetry from timing, memory and performance counters: none on the map
- Hardware-attested weight binding: [Attestable Audits](https://trustbutveri.fyi/implementations/attestable-audits/) (R2, research prototype); [PAL*M](https://trustbutveri.fyi/implementations/palm/) (R2, research prototype); [Tinfoil model identity (Modelwrap)](https://trustbutveri.fyi/implementations/tinfoil-model-identity/) (R3, product)

## Sources

1. Verification Plan, R. Dean (2026). https://ai-2040.com/supplements/verification-plan
2. Example Schemes for Verifying High-Stakes AI Agreements, Amodo Design (2026). https://amododesign.com/notes/2026-06-23-verification-algorithms/
3. Scaling Recomputation Inference Verification, Amodo Design (2026). https://amododesign.com/notes/2026-09-02-scaling-recomputation-inference-verification/
4. AI 2040 Plan A — Verification SITREP, Amodo Design (2026). https://amododesign.com/ai-verification/plan-a-sitrep/
5. Get Involved in Verification, AI Futures Project (2026). https://ai-2040.com/supplements/verification-plan/get-involved
6. Proof-of-Learning is Currently More Broken Than You Think, C. Fang et al. (2023). https://arxiv.org/abs/2208.03567
7. Timing and Memory Telemetry on GPUs for AI Governance, S. K. Monfared et al. (2026). https://arxiv.org/abs/2602.09369
8. Guaranteeable Memory: An HBM-Based Chiplet for Verifiable AI Workloads, J. Petrie (2025). https://openreview.net/forum?id=uc79kOv0MV
9. Detecting Hidden ML Training With Zero-Overhead Telemetry, R. Rahman & S. Tajdari (2026). https://arxiv.org/abs/2606.19262
10. On TEEs for Privacy-Preserving Monitoring in AI Governance, Gloria Z (2026). https://techgov.intelligence.org/blog/on-tees-for-privacy-preserving-monitoring-in-ai-governance
11. NVIDIA Secure AI with Blackwell and Hopper GPUs (White Paper), NVIDIA (2025). https://docs.nvidia.com/nvidia-secure-ai-with-blackwell-and-hopper-gpus-whitepaper.pdf
12. How Tinfoil Proves Exactly What Model Is Running, Tinfoil Team (2026). https://tinfoil.sh/blog/2026-02-03-proving-model-identity
13. PAL*M: Property Attestation for Large Generative Models, P. Chantasantitam et al. (2026). https://arxiv.org/abs/2601.16199
14. A primer on secure enclaves, Tinfoil (2026). https://docs.tinfoil.sh/verification/secure-enclave-primer
15. Backend infrastructure, Tinfoil (2026). https://docs.tinfoil.sh/verification/attestation-architecture
16. How verification works in Tinfoil, Tinfoil (2026). https://docs.tinfoil.sh/verification/verification-in-tinfoil
17. modelwrap: Reproducible dm-verity read-only image of Huggingface models, Tinfoil (2026). https://github.com/tinfoilsh/modelwrap
18. TEE.fail: Breaking Trusted Execution Environments via DDR5 Memory Bus Interposition, J. Chuang et al. (2026). https://tee.fail/
19. Battering RAM: Low-Cost Interposer Attacks on Confidential Computing via Dynamic Memory Aliasing, J. De Meulemeester et al. (2026). https://batteringram.eu/
20. RMPocalypse: How a Catch-22 Breaks AMD SEV-SNP, B. Schlüter & S. Shinde (2025). https://rmpocalypse.github.io/
21. SEV-SNP RMP Initialization Vulnerability (AMD-SB-3020), AMD (2025). https://www.amd.com/en/resources/product-security/bulletin/amd-sb-3020.html
22. Attestable Audits: Verifiable AI Safety Benchmarks Using Trusted Execution Environments, C. Schnabl et al. (2025). https://arxiv.org/abs/2506.23706