# 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-0023,M-0018&hide=training

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.

- **Keep hidden from the verifier: training data.** 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 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 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Safeguard attestation | R2 | Semi-trusted | Analysis | Existing features | 0 / 5 / 0 | depends | depends | not involved |
| Chip location verification | R1 | Adversarial | Analysis | Existing features | 0 / 4 / 0 | not involved | not involved | not involved |

## Claims

No claims chosen.

## Mechanisms

### Safeguard attestation

Hardware-signed evidence that an AI service ran its declared safeguards, such as a guardrail classifier or monitor, when producing a given response. ([Safeguard attestation](https://trustbutveri.fyi/mechanisms/safeguard-attestation/))

- Assessment: mechanism family.
- Readiness: R2 Demonstrated, assessed for attesting that a declared safeguard mediated a service's responses.
- Claims in this proposal: none of them.
- Threat model: semi-trusted prover. Hardware: existing features. Prover cooperation: required. Attack testing: analysis. Category: Cryptographic & computational.
- What the verifier sees: model weights depends; inputs and outputs depends; training data not involved. The enclave route signs hashes of the safeguard, request and response; a low-trust design has the verifier re-run and screen sampled requests itself.

### Chip location verification

Timing a chip's signed replies to trusted servers at known places, so that the speed of light bounds how far away the chip can be. ([Chip location verification](https://trustbutveri.fyi/mechanisms/chip-location-verification/))

- Assessment: mechanism family.
- Readiness: R1 Proposed, assessed for bounding how far a chip is from trusted landmark servers when checked.
- Claims in this proposal: none of them.
- Threat model: adversarial prover. Hardware: existing features. Prover cooperation: required. Attack testing: analysis. Category: Compute accounting & provenance.
- What the verifier sees: model weights not involved; inputs and outputs not involved; training data not involved. Times signed replies from chips; it does not handle model data.


## Properties

**Built for an adversarial prover**

- Chip location verification

**No new hardware needed**

- Safeguard attestation
- Chip location verification


## 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**

- Safeguard attestation: Analysis
- Chip location verification: Analysis


## Limits

**Open significant flaws**

- Attestation shows a safeguard ran, not that it is effective (theoretical argument, in Safeguard attestation; https://trustbutveri.fyi/mechanisms/safeguard-attestation/#flaw-1) [1]. Proof of guardrail ensures that the guardrail executed, but the guardrail can still err or be jailbroken. Because the guardrail must be open source, a malicious developer can attack it with jailbreaks while still presenting a valid proof. In the authors' evaluation, Llama Guard 3 reached an F1 score of 0.56 on the unsafe class of the ToxicChat dataset. The authors state that proof of guardrail should not be interpreted or advertised as proof of safety.
- Selective attestation leaves traffic uncovered (theoretical argument, in Safeguard attestation; https://trustbutveri.fyi/mechanisms/safeguard-attestation/#flaw-2) [1][4][9]. Attestations are issued per response. In the prototype, the agent offers them when it receives high-stakes questions, so nothing shows that unattested traffic went through the same path. PAL*M's authors note that a prover could cherry-pick favourable executions, and suggest verifier-published nonces or requesting only session-level proofs. A governance analysis notes that auditors also need assurance that all activity is accounted for, since a host could start a second confidential virtual machine that bypasses monitoring.

  Related mechanism: On-chip telemetry from timing, memory and performance counters (R2, not in the proposal). On-chip counters are a proposed route to evidence about everything a chip runs, including a second virtual machine that skips the safeguard.
- Measurements may omit behaviour-relevant configuration or runtime changes (theoretical argument, in Safeguard attestation; https://trustbutveri.fyi/mechanisms/safeguard-attestation/#flaw-3) [9]. Every component that influences inference behaviour must be covered by the launch measurement, including feature flags, environment variables and invocation arguments. A launch measurement also does not show that a program keeps running as measured if the kernel is later compromised.
- Components outside the attested boundary (theoretical argument, in Safeguard attestation; https://trustbutveri.fyi/mechanisms/safeguard-attestation/#flaw-4) [1][2]. In the proof-of-guardrail experiments, the guardrail model and the agent's backend model were both reached through external APIs, and the authors leave the decision to trust those APIs to the verifier. The measured wrapper must also have no vulnerability that lets the unmeasured agent bypass the guardrail, for example by executing arbitrary commands inside the enclave. The code's README states that the enclave does not currently restrict the agent's arbitrary command execution, which could be used to bypass guardrails.
- Memory-bus interposition extracts attestation keys and forges attestations (demonstrated attack, in Safeguard attestation; https://trustbutveri.fyi/mechanisms/safeguard-attestation/#flaw-5) [1][4][7][8][9][10][11][12][13]. Inherited finding. Applies to variants using the affected Intel or AMD trust roots. PAL*M excludes physical attacks. A TDX-backed safeguard claim against a physical host attacker would be defeated, but these studies do not demonstrate a break of the AWS Nitro proof-of-guardrail prototype or of verifier-side recomputation. The TEE findings cover DDR5 attacks on Intel TDX, the H100 relay demonstration, DDR4 attacks on AMD SEV-SNP, and software-only SEV-SNP forgery before AMD's fixes. These are inherited hardware limits; a governance analysis explains why physical access matters in a treaty setting. Related finding: https://trustbutveri.fyi/mechanisms/tee-remote-attestation/#flaw-1.

  Response: Intel and AMD place the physical attack class outside their threat models, according to the researchers. AMD reports firmware 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.
- Extracting a chip's key lets another device answer for it (theoretical argument, in Chip location verification; https://trustbutveri.fyi/mechanisms/chip-location-verification/#flaw-1) [14][19]. Ping-based protocols rely on cryptographic keys stored on the chip. Tee and Happel argue that an adversary with physical access could extract these keys and so compromise location verification. They propose GPU fingerprints as a mitigation, so far tested on 24 GPUs. Brass and Aarne assume the keys are stored securely, for example in a TPM.
- Added delay can shift an estimated position (demonstrated attack, in Chip location verification; https://trustbutveri.fyi/mechanisms/chip-location-verification/#flaw-2) [14][17]. Brass and Aarne cite internet-geolocation research in which artificially increased round-trip times moved the estimated location by up to 1,000 km, with a 74% chance of avoiding detection. Avellar and Grunewald list inflated ping times from circuitous routing as an evasion route. Added delay only loosens a distance bound, and Brass and Aarne propose a hard time limit as the counter: a chip that replies too slowly cannot be ruled out of a restricted location.
- Faster-than-assumed network paths (theoretical argument, in Chip location verification; https://trustbutveri.fyi/mechanisms/chip-location-verification/#flaw-3) [14][17]. Brass and Aarne list dark fibre and other private high-speed interconnects as ways to lower measured delays artificially. They judge that leasing dark fibre would probably not be a considerable challenge for covertly or openly adversarial actors. Avellar and Grunewald note that this can make a chip appear to be somewhere else entirely. A limit set at the vacuum speed of light cannot be beaten, but it makes honest chips fail more often.
- Compromised landmarks can falsify measurements (theoretical argument, in Chip location verification; https://trustbutveri.fyi/mechanisms/chip-location-verification/#flaw-4) [14][17][18]. A party that controls landmark servers can report false timing. Brass and Aarne cite research in which manipulating a third of the landmarks shifted the estimated location by about 700 km. Avellar and Grunewald note that compromised landmarks let adversaries spoof travel-time measurements directly. The draft specification asks verifiers to require anchors in diverse places, run by several independent operators.

**Not yet demonstrated**

- Chip location verification: R1 Proposed, assessed for bounding how far a chip is from trusted landmark servers when checked


## 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.

- **On-chip telemetry from timing, memory and performance counters** (R2 Demonstrated, assessed for workload evidence from GPU counters and timing, assuming authentic measurements)
  - Bears on the open significant flaw "Selective attestation leaves traffic uncovered" in Safeguard attestation. On-chip counters are a proposed route to evidence about everything a chip runs, including a second virtual machine that skips the safeguard.
- **Hardware-enabled guarantees (flexHEG) and guarantee processors** (R1 Proposed, assessed for checking and enforcing training-compute limits on chips, against adversaries up to states)
  - Bears on the open significant flaw "Memory-bus interposition extracts attestation keys and forges attestations" in Safeguard attestation. A tamper-protected enclosure around the chip is the proposed answer when the party that holds the hardware may attack it physically.
- **Model identity attestation** (R3 In production, assessed for showing users that a service runs the declared model weights)
  - Safeguard attestation waits on it: Safeguard evidence must be bound to the model actually served, which depends on model-identity attestation.
- **TEE remote attestation for AI workloads** (R3 In production, assessed for showing which software ran to a party that distrusts the operator holding the hardware)
  - Safeguard attestation waits on it: Frontier model inference typically needs several GPUs, GPU confidential computing is less mature than CPU support, and CPU inference, which an enclave prototype had to use, ran about 100 times slower than GPU inference.
  - Safeguard attestation waits on it: Trust rests on a small number of hardware vendors, and a per-CPU Intel attestation key has been extracted by physical attack.
  - Chip location verification depends on it.


## Dependencies

**Missing prerequisites**

- TEE remote attestation for AI workloads (R3 In production, assessed for showing which software ran to a party that distrusts the operator holding the hardware), needed by Safeguard attestation and Chip location verification
- Model identity attestation (R3 In production, assessed for showing users that a service runs the declared model weights), needed by Safeguard attestation

**Shared foundations**

- TEE remote attestation for AI workloads, relied on by Safeguard attestation and Chip location verification

**Blockers**

- Safeguard attestation: No published design shows that all of a provider's traffic passes through the attested safeguard path; current evidence covers individual attested responses. (coverage & hidden compute) [1][9]
- Safeguard attestation: Frontier model inference typically needs several GPUs, GPU confidential computing is less mature than CPU support, and CPU inference, which an enclave prototype had to use, ran about 100 times slower than GPU inference. (performance & compatibility; waits on TEE remote attestation for AI workloads) [9][21]
- Safeguard attestation: Trust rests on a small number of hardware vendors, and a per-CPU Intel attestation key has been extracted by physical attack. (hardware trust; waits on TEE remote attestation for AI workloads) [7][9]
- Safeguard attestation: Safeguard evidence must be bound to the model actually served, which depends on model-identity attestation. (evidence binding; waits on Model identity attestation) [21][22]
- Safeguard attestation: No independent red-team or audit of a safeguard-attestation system has been published, and the available prototypes are described by their authors as proofs of concept that have not been stress-tested by a counterparty. (adversarial validation) [2][6]
- Chip location verification: No public code or reproducible end-to-end location results are available for the reported H100 prototype. (adversarial validation) [15][16]
- Chip location verification: Per-chip keys must be provisioned and protected against extraction; hardware-integrated, tamper-resistant versions still need R&D. (hardware trust) [14][19][23]
- Chip location verification: The time limit forces a trade-off: a limit at the speed of light in fibre can be beaten by faster links, while one at the vacuum speed of light makes honest chips fail often. (protocol soundness) [14]
- Chip location verification: A trusted landmark network must be built and secured, and who should operate it, under what oversight, is unsettled. (access & governance) [14][17]


## What the verifier sees

- Model weights: shown by none; depends on the design for Safeguard attestation; hidden by none; not involved in Chip location verification; unspecified for none.
- Inputs and outputs: shown by none; depends on the design for Safeguard attestation; hidden by none; not involved in Chip location verification; unspecified for none.
- Training data: shown by none; depends on the design for none; hidden by none; not involved in Safeguard attestation and Chip location verification; unspecified for none.

## Implementations

- Safeguard attestation: none on the map
- Chip location verification: [Lucid sovereignty (location) certificates](https://trustbutveri.fyi/implementations/lucid-location-certificates/) (R1, standard)

## Sources

1. Proof-of-Guardrail in AI Agents and What (Not) to Trust from It, X. Jin et al. (2026). https://arxiv.org/abs/2603.05786
2. Verifiable-ClawGuard: proof-of-guardrail reference code, SaharaLabsAI (2026). https://github.com/SaharaLabsAI/Verifiable-ClawGuard
3. Safety Without Compromising on Privacy, D. McCann-Sayles et al. (2026). https://tinfoil.sh/blog/2026-09-14-safety-without-compromising-privacy
4. PAL*M: Property Attestation for Large Generative Models, P. Chantasantitam et al. (2026). https://arxiv.org/abs/2601.16199
5. Enabling Verifiably-Scoped Monitoring through Large Language Models and Trusted Compute, B. Penchas et al. (2026). https://icml.cc/virtual/2026/78630
6. Auditor-in-a-Box: Tools for Third-Party Auditing, R. Rinberg & B. Penchas (2026). https://www.lesswrong.com/posts/uWYk7MM9hAf9GEbGe/auditor-in-a-box-tools-for-third-party-auditing
7. TEE.fail: Breaking Trusted Execution Environments via DDR5 Memory Bus Interposition, J. Chuang et al. (2026). https://tee.fail/
8. DDRop: Active Memory Interposer Attacks on Confidential VMs by Dropping DDR5 Writes, J. De Meulemeester et al. (2026). https://ddropattack.eu/
9. 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
10. Battering RAM: Low-Cost Interposer Attacks on Confidential Computing via Dynamic Memory Aliasing, J. De Meulemeester et al. (2026). https://batteringram.eu/
11. RMPocalypse: How a Catch-22 Breaks AMD SEV-SNP, B. Schlüter & S. Shinde (2025). https://rmpocalypse.github.io/
12. SEV-SNP RMP Initialization Vulnerability (AMD-SB-3020), AMD (2025). https://www.amd.com/en/resources/product-security/bulletin/amd-sb-3020.html
13. 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
14. Location Verification for AI Chips, A. Brass & O. Aarne (2024). https://www.iaps.ai/research/location-verification-for-ai-chips
15. Location Verification for AI Chips (issue brief), A. Brass (2025). https://static1.squarespace.com/static/64edf8e7f2b10d716b5ba0e1/t/6827b67275666f3757f134ea/1747433075281/Location+Verification+two-pager.pdf
16. Ping-based Location, Ulyssean (2025). https://ping-location.info/
17. Near-Term Verification Methods for AI Chip Exports, B. Avellar & E. Grunewald (2026). https://www.iaps.ai/research/near-term-verification-methods-for-ai-chip-exports
18. Sovereignty Certificates: draft specification, version 0.1.0, Sovereignty Certificates Working Group (2025). https://github.com/Lucid-Computing/sovereignty-certificate-specification
19. GPU Fingerprinting for Location Verification, W. Tee & J. Happel (2026). https://arxiv.org/abs/2605.01930
20. Secure, Governable Chips: Using On-Chip Mechanisms to Manage National Security Risks from AI & Advanced Computing, O. Aarne et al. (2024). https://www.cnas.org/publications/reports/secure-governable-chips
21. Attestable Audits: Verifiable AI Safety Benchmarks Using Trusted Execution Environments, C. Schnabl et al. (2025). https://arxiv.org/abs/2506.23706
22. How Tinfoil Proves Exactly What Model Is Running, Tinfoil Team (2026). https://tinfoil.sh/blog/2026-02-03-proving-model-identity
23. Hardware-Level Governance of AI Compute: A Feasibility Taxonomy for Regulatory Compliance and Treaty Verification, S. Ansari (2026). https://arxiv.org/abs/2604.04712