# AI verification proposal

A proposal built with the Proposal Explorer of the AI Verification Tech Map (https://trustbutveri.fyi/), from its records of 2026-10-08. Interactive version: https://trustbutveri.fyi/explorer/?mechanisms=M-0004,M-0016&implementations=M-0004:I-0005

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 readiness and open flaws. Readiness levels R0 to R4 describe one record's public evidence for its assessed use and are never combined. Definitions: https://trustbutveri.fyi/about/methodology/ (roles, properties and flaws) and https://trustbutveri.fyi/about/readiness/ (readiness levels).

## Filters

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

None set. Every mechanism on the map was available.

## Overview

One row per mechanism, read from its record. Open flaws: critical / significant / minor. The last three columns are the editors' reading of what the verifier sees.

| Mechanism | Readiness | Prover | Attack testing | Hardware | Open flaws | Weights | Inputs and outputs | Training data |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Zero-knowledge proofs of inference / Attestable zero-knowledge inference prover | R1 | Adversarial | None | None | 0 / 2 / 0 | unspecified | unspecified | unspecified |
| Timed challenge-response and memory-occupation challenges | R2 | Adversarial | Analysis | None | 0 / 1 / 1 | not involved | not involved | not involved |

## Claims

No claims chosen.

## Mechanisms

### Zero-knowledge proofs of inference

Attestable's zero-knowledge prover, which the company reports proves large language model outputs came from committed weights at tens of tokens per second. ([Zero-knowledge proofs of inference](https://trustbutveri.fyi/mechanisms/zk-proofs-of-inference/))

- Assessment: selected implementation [Attestable zero-knowledge inference prover](https://trustbutveri.fyi/implementations/attestable-zk-inference/).
- Readiness: R1 Proposed, assessed for proving an output came from committed weights.
- Claims in this proposal: none of them.
- Threat model: adversarial prover. Hardware: none. Prover cooperation: required. Attack testing: none. Category: Cryptographic & computational.
- What the verifier sees: model weights unspecified; inputs and outputs unspecified; training data unspecified. This Explorer has no asset-specific exposure assessment for this implementation. Check its source and deployment assumptions.

### Timed challenge-response and memory-occupation challenges

A verifier sends unpredictable questions that a device can answer in time only if it holds specified data, or dedicates specified resources, locally. ([Timed challenge-response and memory-occupation challenges](https://trustbutveri.fyi/mechanisms/timed-challenge-response/))

- Assessment: mechanism family.
- Readiness: R2 Demonstrated, assessed for detecting whether a GPU is doing other work.
- Claims in this proposal: none of them.
- Threat model: adversarial prover. Hardware: none. Prover cooperation: required. Attack testing: analysis. Category: Cryptographic & computational.
- What the verifier sees: model weights not involved; inputs and outputs not involved; training data not involved. Uses verifier-chosen challenges; it does not handle model data.


## Properties

**Built for an adversarial prover**

- Zero-knowledge proofs of inference
- Timed challenge-response and memory-occupation challenges

**No new hardware needed**

- Zero-knowledge proofs of inference
- Timed challenge-response and memory-occupation challenges


## Attack testing

Published attempts to break a system, including those that found failures. Testing history does not establish that open flaws are resolved.

**Testing history**

- Timed challenge-response and memory-occupation challenges: Analysis


## Limits

**Open significant flaws**

- Proves an 8-bit quantised variant of the model (open question, in Attestable zero-knowledge inference prover; https://trustbutveri.fyi/implementations/attestable-zk-inference/#flaw-1) [1]. Attestable reports that matrix multiplications are dynamically quantised to 8-bit integers, while non-linear operations are proven in floating point. It reports that its IFEval result "shows where the current quantization still needs improvement". The proven model is therefore a quantised variant, which a verifier must accept as the declared model.
- A proof covers only the computation it is about (theoretical argument, in Attestable zero-knowledge inference prover; https://trustbutveri.fyi/implementations/attestable-zk-inference/#flaw-2) [2]. Attestable states that "a proof of some computation is not a proof of all computation" and that a proof "cannot discover a datacenter that was never declared".
- Remote memory narrows the timing margin (theoretical argument, in Timed challenge-response and memory-occupation challenges; https://trustbutveri.fyi/mechanisms/timed-challenge-response/#flaw-2) [6]. Data-centre remote memory access returns in about 1–2 µs, against about 70–200 ns for local DRAM. The MIRI overview says verification of memory saturation depends on ruling out remote access by latency or physical disconnection. It adds that pre-staging data is ruled out only by unpredictable, capacity-filling challenges.

  Related mechanism: Bandwidth limits and compartmentalization (R2, not in the proposal). Physical disconnection is proposed to exclude remote memory between the separated groups during a challenge. It depends on the isolation boundary being enforced.

**Family finding context**

- Context for Attestable zero-knowledge inference prover; applicability depends on the finding's scope. The proof covers a fixed-point approximation, not the floating-point model (open question, in Zero-knowledge proofs of inference; https://trustbutveri.fyi/mechanisms/zk-proofs-of-inference/#flaw-1) [1][4][5][6]. Current ZK inference systems prove a quantised version of the network. zkLLM scales values by 2^16 and reports small perplexity changes. Attestable reports quantising matrix multiplications to 8-bit integers while proving other operations in floating point. A verifier therefore learns about the proof-friendly variant, and must separately accept that this variant is the declared model. Trail of Bits built a ResNet-18 backdoor that is dormant in the full-precision model and active after ezkl's quantisation; whether it persists through proving was left for further investigation. A verification system design calls floating-point emulation in ZKPs an open problem.
- Context for Attestable zero-knowledge inference prover; applicability depends on the finding's scope. A proof speaks only for the computations that were proven (theoretical argument, in Zero-knowledge proofs of inference; https://trustbutveri.fyi/mechanisms/zk-proofs-of-inference/#flaw-2) [2]. Attestable writes that "a proof of some computation is not a proof of all computation", and that a proof cannot discover a datacenter that was never declared. Proofs of inference do not by themselves show that no other workload ran on the same or other hardware.

  Related mechanism: Proofs of useful work for capacity accounting (R1, not in the proposal). The record names proof-of-work accounting as the kind of compute accounting needed to show that proven inference was the only work done.
- Context for Attestable zero-knowledge inference prover; applicability depends on the finding's scope. The model architecture is disclosed (theoretical argument, in Zero-knowledge proofs of inference; https://trustbutveri.fyi/mechanisms/zk-proofs-of-inference/#flaw-3) [4][7]. ZKML "requires that the model architecture (but not weights) is revealed", and zkLLM assumes a publicly known model structure. Architecture can be commercially sensitive.
- Context for Attestable zero-knowledge inference prover; applicability depends on the finding's scope. Proofs do not bind computational effort (Hollow-LLM) (demonstrated attack, in Zero-knowledge proofs of inference; https://trustbutveri.fyi/mechanisms/zk-proofs-of-inference/#flaw-4) [8]. Researchers at the University of Southern California show that a proof of inference certifies that an output is consistent with committed weights under the declared architecture, but not how much computation produced it. In their Hollow-LLM attack, a provider keeps the declared architecture and parameter count but commits to "ghost weights". Some layers pass their inputs through unchanged, and wide layers carry the signal in a small subspace, so a much smaller inner model does the real work. The ghost weights satisfy the verification circuit and yield valid proofs.

  The authors ran the attack with the proof procedure of zkGPT, a separate ZK inference system, on a 6-layer, 512-dimensional transformer declared as up to 12 layers and 1,024 dimensions. Outputs were identical to the inner model's, and serving cost stayed at the inner model's level. An honest model of the declared size cost 2.4 times as much to prefill and 3.1 times as much to decode. Proving cost still grew with the declared architecture.

  The authors note that results may be served before any proof, with the provider building the witness only when a call is selected for audit. They describe their constructions as "compatible with state-of-the-art zkLLM pipelines", and state that the attack does not imply a flaw in the proof system itself. They propose challenge-based audits and ablation tests, which raise the cost of cheating but give no guarantee.

**Open minor flaws**

- Error rates not quantified (open question, in Timed challenge-response and memory-occupation challenges; https://trustbutveri.fyi/mechanisms/timed-challenge-response/#flaw-3) [10]. Monfared et al. show separable timing distributions but do not define thresholds or statistical tests, so false-positive and false-negative rates are not quantified.

**Not yet demonstrated**

- Zero-knowledge proofs of inference: R1 Proposed, assessed for proving an output came from committed weights


## Possible additions

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

- **Bandwidth limits and compartmentalization** (R2 Demonstrated, assessed for monitoring inter-node traffic with operator-run software on four GPUs)
  - Bears on the open significant flaw "Remote memory narrows the timing margin" in Timed challenge-response and memory-occupation challenges. Physical disconnection is proposed to exclude remote memory between the separated groups during a challenge. It depends on the isolation boundary being enforced.
  - Timed challenge-response and memory-occupation challenges waits on it: Outside help, such as remote memory, must be excluded during challenges.
- **Proofs of useful work for capacity accounting** (R1 Proposed, assessed for bounding the spare capacity of declared hardware that could run training)
  - Zero-knowledge proofs of inference waits on it: Covering computation that is not proven relies on proof-of-work accounting, which Attestable has only proposed.


## Dependencies

**Blockers**

- Zero-knowledge proofs of inference: No paper, protocol specification or code is public, so the reported results cannot be reproduced. (adversarial validation) [1]
- Zero-knowledge proofs of inference: Attestable reports a context window limited to 16K tokens. (performance & compatibility) [1]
- Zero-knowledge proofs of inference: Covering computation that is not proven relies on proof-of-work accounting, which Attestable has only proposed. (coverage & hidden compute; waits on Proofs of useful work for capacity accounting) [2]
- Timed challenge-response and memory-occupation challenges: No network-level memory challenge across data-centre servers has been demonstrated. (adversarial validation) [6]
- Timed challenge-response and memory-occupation challenges: Challenges that fill memory displace workloads; filling a pod's volatile memory takes tens of minutes and SSDs take hours. (performance & compatibility) [6][10]
- Timed challenge-response and memory-occupation challenges: Outside help, such as remote memory, must be excluded during challenges. (coverage & hidden compute; waits on Bandwidth limits and compartmentalization) [6]


## What the verifier sees

- Model weights: shown by none; depends on the design for none; hidden by none; not involved in Timed challenge-response and memory-occupation challenges; unspecified for Zero-knowledge proofs of inference.
- Inputs and outputs: shown by none; depends on the design for none; hidden by none; not involved in Timed challenge-response and memory-occupation challenges; unspecified for Zero-knowledge proofs of inference.
- Training data: shown by none; depends on the design for none; hidden by none; not involved in Timed challenge-response and memory-occupation challenges; unspecified for Zero-knowledge proofs of inference.

## Implementations

- Zero-knowledge proofs of inference: [Attestable zero-knowledge inference prover](https://trustbutveri.fyi/implementations/attestable-zk-inference/) (R1, product); [EZKL](https://trustbutveri.fyi/implementations/ezkl/) (R2, product); [Low-trust AI compute verification system overview](https://trustbutveri.fyi/implementations/low-trust-compute-verification-system-overview/) (R1, proposed architecture); [zkLLM](https://trustbutveri.fyi/implementations/zkllm/) (R2, research prototype)
- Timed challenge-response and memory-occupation challenges: [Data-centre memory challenging](https://trustbutveri.fyi/implementations/data-centre-memory-challenging/) (R1, proposed architecture); [GPU contention probes](https://trustbutveri.fyi/implementations/gpu-contention-probes/) (R2, research prototype); [Low-trust AI compute verification system overview](https://trustbutveri.fyi/implementations/low-trust-compute-verification-system-overview/) (R1, proposed architecture); [SAGE](https://trustbutveri.fyi/implementations/sage-gpu-attestation/) (R2, research prototype); [VRAM-residency challenge](https://trustbutveri.fyi/implementations/vram-residency-challenge/) (R2, research prototype)

## Sources

1. Proving LLMs at Scale, Attestable (2026). https://attestable.com/blog/proving-llms-scale
2. Pacing AI Requires Proof, Attestable (2026). https://attestable.com/blog/pacing-ai-requires-proof
3. From Verifiability to Model-Weight Security, Attestable (2026). https://attestable.com/blog/model-weights-security
4. zkLLM: Zero Knowledge Proofs for Large Language Models, H. Sun et al. (2024). https://doi.org/10.1145/3658644.3670334
5. Zkonduit EZKL Security Assessment, F. Casal et al. (2025). https://github.com/trailofbits/publications/blob/master/reviews/2025-03-zkonduit-ezkl-securityreview.pdf
6. A System Overview for Near-Term, Low-Trust AI Compute Verification, N. Cankaya (2026). https://intelligence.org/wp-content/uploads/2026/06/A-system-overview-for-near-term-low-trust-AI-compute-verification.pdf
7. ZKML: An Optimizing System for ML Inference in Zero-Knowledge Proofs, B.-J. Chen et al. (2024). https://doi.org/10.1145/3627703.3650088
8. Hollow-LLM Attack: Computationally Trivial Weights in Zero-Knowledge Verification of LLM Inference, C. Gong et al. (2026). https://arxiv.org/abs/2607.28884
9. Verification Plan, R. Dean (2026). https://ai-2040.com/supplements/verification-plan
10. Timing and Memory Telemetry on GPUs for AI Governance, S. K. Monfared et al. (2026). https://arxiv.org/abs/2602.09369
11. SAGE: Software-based Attestation for GPU Execution, A. Ivanov et al. (2023). https://www.usenix.org/conference/atc23/presentation/ivanov
12. SWATT: SoftWare-based ATTestation for Embedded Devices, A. Seshadri et al. (2004). https://netsec.ethz.ch/publications/papers/swatt.pdf
13. Proofs of Space, S. Dziembowski et al. (2015). https://eprint.iacr.org/2013/796
14. Secure Code Update for Embedded Devices via Proofs of Secure Erasure, D. Perito & G. Tsudik (2010). https://link.springer.com/chapter/10.1007/978-3-642-15497-3_39
15. Software-Based Memory Erasure with Relaxed Isolation Requirements, S. Bursuc et al. (2024). https://ieeexplore.ieee.org/document/10664348/
16. On the Difficulty of Software-Based Attestation of Embedded Devices, C. Castelluccia et al. (2009). https://s3.eurecom.fr/docs/ccs09_Castelluccia.pdf
17. Refutation of "On the Difficulty of Software-Based Attestation of Embedded Devices", A. Perrig & L. van Doorn (2010). https://netsec.ethz.ch/publications/papers/perrig-ccs-refutation.pdf