Proposal Explorer

Add claims to see which ones the mechanisms address.

In production

• R3 Sampled inference recomputation for checking untrusted workers' activations against the declared model, prompt and precision

Built for an adversarial prover

Sampled inference recomputation

No new hardware needed

Sampled inference recomputation

Not counted

Excluded by your filters: Bounding unexplained information in outputs and Bandwidth limits and compartmentalization

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

• Bounding unexplained information in outputs: Independent red-team · excluded by filters
• Sampled inference recomputation: Analysis
• Bandwidth limits and compartmentalization: Analysis · excluded by filters

Excluded by your filters

• Bounding unexplained information in outputs Readiness R2 is below the minimum of R3.
• Bandwidth limits and compartmentalization Readiness R2 is below the minimum of R3.

Open significant flaws

12 flaws in 3 mechanisms

• Prompt-controlled entropy inflation widens the covert channel in Bounding unexplained information in outputs

  Gumbel-based inference verification tolerates token choices that honest GPU nondeterminism could produce, and the size of that tolerated set grows with the model's output entropy. Kezins, an independent researcher, showed that an adversary who controls the prompt distribution can raise output entropy and roughly double the bits leaked per token. Across six models of 1 to 32 billion parameters, this cut the slowdown from 146–254 times under benign prompts to 60–118 times. Kezins argues that architectures built on the same unexplained-information bound inherit this attack surface, and recommends calibrating tolerances against local token entropy rather than benign traffic. ^{2 3}

  Demonstrated attack · Significant · Open. On the record

  Related mechanism R3 Deterministic and bit-exact inference: Bit-exact replay would remove the tolerance for numerical noise that sets the size of this channel. The record notes that it needs full hardware and software metadata. A pointer, not evidence that this flaw is mitigated. Add

• Information the declared computation explains is not bounded in Bounding unexplained information in outputs

  The bound limits unexplained bits only. Outputs that the declared computation fully explains can still carry valuable information: a compression study notes that an adversary with inference access can extract more proprietary information per bit than naive transmission allows. ^{1 4}

  Theoretical argument · Significant · Open. On the record

• Channels other than checked outputs are outside the bound in Bounding unexplained information in outputs

  The inference-verification scheme treats side channels as out of scope. A low-trust system design argues that suppressing physical covert bandwidth below kilobits per second is much more achievable than aiming for zero, and that a malicious device can leak one bit of information by deliberately outputting a wrong result. ^{2 5}

  Theoretical argument · Significant · Open. On the record

  Related mechanism R1 Side-channel suppression for isolated facilities: Physical side channels need separate suppression, which is this mechanism's purpose. A pointer, not evidence that this flaw is mitigated. Add

• The facility-level design is untested in Bounding unexplained information in outputs

  The compute-verification architecture is described with protocol details, potential attacks and prototyping plans, but no prototype results have been published. ¹

  Open question · Significant · Open. On the record

• Tolerance for numerical noise leaves a covert channel in Sampled inference recomputation

  Schemes that accept approximate matches can put an upper bound on an adversary's covert bandwidth, but they cannot close the channel. The weight-exfiltration detector cut exfiltratable information to under 0.5%, not to zero, on a 30-billion-parameter mixture-of-experts model under benign prompt traffic. Its authors called the channel's size under adversarial prompts an open empirical question. An independent study showed that an adversary who controls the prompts roughly doubles the bits leaked per token. Across six models, that cut the slowdown from 146–254 times under benign prompts to 60–118 times. The attack widens the exfiltration bound. It does not target the check that outputs match the declared model. ^{2 3 15}

  Demonstrated attack · Significant · Open. On the record

  Related mechanism R3 Deterministic and bit-exact inference: Bit-exact inference would remove the numerical tolerance if exact replay can be deployed with the required weights and configuration. A pointer, not evidence that this flaw is mitigated. Add

• Only recorded traffic is checked in Sampled inference recomputation

  Recomputation checks that recorded, declared workloads are correct. It cannot show that the record is complete. The published schemes do not cover hidden workloads run on the same compute, or substituted work. Rinberg et al. say their exfiltration-detection scheme cannot stand alone. ^{2 16}

  Theoretical argument · Significant · Open. On the record

  Related mechanism R1 Network taps and certifiers: Taps copy and hash all traffic on the monitored links, which bears on whether the traffic record is complete. They do not show what else ran on the same chips. A pointer, not evidence that this flaw is mitigated. Add

• Some inference optimizations are not covered in Sampled inference recomputation

  TOPLOC's authors state that it cannot detect speculative decoding in which a cheaper model does the decoding. They did not test whether it distinguishes types of key-value (KV) cache compression. DiFR was evaluated only on sampling from a single model. Its authors sketch an extension to one speculative-decoding algorithm but do not test it. ^{6 7}

  Theoretical argument · Significant · Open. On the record

• Low-communication training reduces the bandwidth training needs in Bandwidth limits and compartmentalization

  DiLoCo matched fully synchronous training on 8 workers while communicating 500 times less. Rahman writes that this family of methods theoretically allows large-scale training with less than 100 Mbps. Lucid includes these methods in its bounds, but notes that extreme activation compression, architectures with unusually small inter-layer widths, or modular paradigms could erode the margin. ^{18 21 22}

  Theoretical argument · Significant · Open. On the record

• Operator control of pod routing collapses the bound in Bandwidth limits and compartmentalization

  Lucid's analysis finds that if the operator can freely assign pods to routers, it could dedicate a whole cell of 100 or more pods to one pipeline stage. The bound then falls to about 90–220x uncompressed and as low as about 25x with compression. The proposed mitigation, auditor-controlled random assignment that is periodically re-randomized, has not been implemented. ¹⁸

  Theoretical argument · Significant · Open. On the record

• Undeclared local storage raises per-pod capacity in Bandwidth limits and compartmentalization

  More memory or storage per pod helps an adversary. Lucid requires per-pod storage to be declared, capped and physically inspected. ¹⁸

  Theoretical argument · Significant · Open. On the record

• Training within one pod is not covered in Bandwidth limits and compartmentalization

  Lucid's bounds concern pre-training models larger than the pods are sized for. Training models that fit in one pod, fine-tuning and reinforcement-learning post-training within one pod are outside the modelled threat. ¹⁸

  Open question · Significant · Open. On the record

• Parallel scale-up switches are hard enforcement points in Bandwidth limits and compartmentalization

  In GB200 topologies, GPUs reach GPUs in other nodes through NVSwitches without a NIC on the path. Amodo notes that limits are hard to enforce there because many switches work in parallel, so compromising one or two would bypass the limit. ²⁰

  Theoretical argument · Significant · Open. On the record

Open minor flaws

1 mechanism with minor findings

• Mixed hardware widens the honest baseline in Sampled inference recomputation

  When honest reference runs span different GPU types, the spread of benign scores grows. In DiFR's tests on Qwen3-30B-A3B, pooling A100 and H200 runs left Token-DiFR unable to separate the two smallest tested changes, a temperature of 1.1 instead of 1.0 and a simulated top-2 sampling bug, at the target false-positive rate, while cross-entropy separated them. Matched provider and verifier environments, or pooling that weights rare large deviations, restored detection. ⁶

  Open question · Minor · Open. On the record

Mechanisms on the map that are not in the proposal. Pointers, not recommendations. ?

• R3 Deterministic and bit-exact inference Add
  • Bears on the open significant flaw “Prompt-controlled entropy inflation widens the covert channel” in Bounding unexplained information in outputs. Bit-exact replay would remove the tolerance for numerical noise that sets the size of this channel. The record notes that it needs full hardware and software metadata.
  • Bears on the open significant flaw “Tolerance for numerical noise leaves a covert channel” in Sampled inference recomputation. Bit-exact inference would remove the numerical tolerance if exact replay can be deployed with the required weights and configuration.
  • Bounding unexplained information in outputs waits on it. Tolerance for numerical nondeterminism sets the size of the residual channel; bit-exact replay would remove it but needs full hardware and software metadata.
• R1 Network taps and certifiers Add

  Excluded by filters: readiness R1

  • Bears on the open significant flaw “Only recorded traffic is checked” in Sampled inference recomputation. Taps copy and hash all traffic on the monitored links, which bears on whether the traffic record is complete. They do not show what else ran on the same chips.
  • Sampled inference recomputation waits on it. In tap-based retrofit designs, recording all inference traffic needs network taps and recomputation servers that can ingest it, in the worst case one recomputation-server network interface per inference front-end interface.
  • Bandwidth limits and compartmentalization waits on it. The verifier must know that all traffic leaving a pod crosses the capped, monitored links.
  • Bounding unexplained information in outputs depends on it.
• R1 Side-channel suppression for isolated facilities Add

  Excluded by filters: readiness R1

  • Bears on the open significant flaw “Channels other than checked outputs are outside the bound” in Bounding unexplained information in outputs. Physical side channels need separate suppression, which is this mechanism's purpose.
  • Bounding unexplained information in outputs waits on it. Physical side channels need separate suppression, and one design treats a low residual bandwidth, rather than zero, as the realistic target.
• R2 Tamper evidence for verifier devices Add

  Excluded by filters: readiness R2

  • Bandwidth limits and compartmentalization waits on it. Shaping devices and routing assignments must be trusted by both parties; Amodo has not yet fully analysed resilience to a compromised DPU.

Missing prerequisites

• R1 Side-channel suppression for isolated facilities needed by Bounding unexplained information in outputs Add
• R1 Network taps and certifiers needed by Bounding unexplained information in outputs Add
• R2 Tamper evidence for verifier devices needed by Bandwidth limits and compartmentalization Add

Shared foundations

• Network taps and certifiers relied on by Bounding unexplained information in outputs, Sampled inference recomputation and Bandwidth limits and compartmentalization

Blockers

14 blockers recorded

• Bounding unexplained information in outputs
  • The prover's compute must be isolated so that all traffic passes through the verifier's interlock; any unmonitored path voids the bound. Coverage & hidden compute. Waits on Bandwidth limits and compartmentalization ¹
  • Physical side channels need separate suppression, and one design treats a low residual bandwidth, rather than zero, as the realistic target. Coverage & hidden compute. Waits on Side-channel suppression for isolated facilities ^{2 5}
  • Tolerance for numerical nondeterminism sets the size of the residual channel; bit-exact replay would remove it but needs full hardware and software metadata. Protocol soundness. Waits on Deterministic and bit-exact inference ^{3 5}
  • Recomputation over confidential weights and inputs needs a protected setting: prover recomputation in a verifier-controlled enclosure, verifier recomputation in a prover-controlled enclosure, or zero-knowledge proofs. Privacy & leakage ¹
  • No prototype of the facility-level architecture exists to red-team. Adversarial validation ¹
• Sampled inference recomputation
  • In tap-based retrofit designs, recording all inference traffic needs network taps and recomputation servers that can ingest it, in the worst case one recomputation-server network interface per inference front-end interface. Coverage & hidden compute. Waits on Network taps and certifiers ^{12 17}
  • In retrofit designs, the recomputation server must sit inside the prover's data centre, possibly under the prover's physical control, and still be protected from a compromised provider, which Amodo rates 'not on track'. Hardware trust ^{2 12}
  • No independent red-team of a recomputation consistency check has been published (the one independent attack study targets the weight-exfiltration bound), and Amodo rates recomputation red-teaming 'not started'. Adversarial validation ^{3 12}
  • Tolerance-based checks need calibration on trusted hardware and exact knowledge of the provider's sampling procedure, and in one prototype a sampling-implementation mismatch produced large spurious differences. Performance & compatibility ^{6 11}
  • The verifier needs the model weights, so checking a closed-weights model requires a trusted, confidential recomputation environment, which the retrofit designs place inside the prover's facility. Privacy & leakage ^{5 6 16}
• Bandwidth limits and compartmentalization
  • No cap that a verifier can check has been implemented or red-teamed. Adversarial validation ¹⁸
  • The verifier must know that all traffic leaving a pod crosses the capped, monitored links. Coverage & hidden compute. Waits on Network taps and certifiers ⁵
  • Shaping devices and routing assignments must be trusted by both parties; Amodo has not yet fully analysed resilience to a compromised DPU. Hardware trust. Waits on Tamper evidence for verifier devices ^{18 20}
  • Advances in low-communication training could shrink the margin that the cap enforces. Capacity bounds ^{18 21 22}

From the family or selected implementation's record. ?

Model weights

Depends on the design for Bounding unexplained information in outputs and Sampled inference recomputation.

Not involved: Bandwidth limits and compartmentalization.

Inputs and outputs

Depends on the design for Bounding unexplained information in outputs and Sampled inference recomputation.

Not involved: Bandwidth limits and compartmentalization.

Training data

Not involved: Bounding unexplained information in outputs, Sampled inference recomputation and Bandwidth limits and compartmentalization.

Bounding unexplained information in outputs

None on the map

Sampled inference recomputation

• R1 AI 2040 inference-only verification stack Proposed architecture, AI Futures Project
• R2 DiFR (Divergence From Reference) Research prototype
• R1 Low-trust AI compute verification system overview Proposed architecture, Machine Intelligence Research Institute
• R1 SASH confidential network logger Research prototype, Singapore AI Safety Hub (SASH)
• R3 TOPLOC Open-source project, Prime Intellect

Bandwidth limits and compartmentalization

• R1 AI 2040 inference-only verification stack Proposed architecture, AI Futures Project
• R1 RAND secure inference data center (SIDC) design Proposed architecture, RAND

1. Verifying AI Compute by Bounding Unexplained Information Exfiltration, J. Petrie & Y. Mühlhäuser (2026). Original
2. Verifying LLM Inference to Detect Model Weight Exfiltration, R. Rinberg et al. (2025). Original
3. Adversarial Entropy Inflation Against Gumbel-Based Inference Verification, N. Kezins (2026). Original
4. Haiku to Opus in Just 10 bits: LLMs Unlock Large Compression Gains, R. Rinberg et al. (2026). Original
5. A System Overview for Near-Term, Low-Trust AI Compute Verification, N. Cankaya (2026). Original
6. DiFR: Inference Verification Despite Nondeterminism, A. Karvonen et al. (2025). Original
7. TOPLOC: A Locality Sensitive Hashing Scheme for Trustless Verifiable Inference, J. M. Ong et al. (2025). Original
8. PrimeIntellect-ai/toploc (GitHub repository), Prime Intellect (2025). Original
9. INTELLECT-2: A Reasoning Model Trained Through Globally Decentralized Reinforcement Learning, Prime Intellect Team et al. (2025). Original
10. adamkarvonen/difr (GitHub repository), A. Karvonen (2025). Original
11. Scaling Recomputation Inference Verification, Amodo Design (2026). Original
12. AI 2040 Plan A — Verification SITREP, Amodo Design (2026). Original
13. SYNTHETIC-2 Release: Four Million Collaboratively Generated Reasoning Traces, Prime Intellect (2025). Original
14. An Inference Verification Prototype — Stage 1, Amodo Design (2026). Original
15. Bit-Exact AI Inference Verification Without Performance Tradeoffs, N. Cankaya (2026). Original
16. Example Schemes for Verifying High-Stakes AI Agreements, Amodo Design (2026). Original
17. Verification Plan, R. Dean (2026). Original
18. Traffic Shaping for Workload Classification, Lucid Computing (2026). Original
19. De-risking Interconnect Limits for AI Verification, A. Scher et al. (2026). Original
20. The Tray as a Bandwidth Boundary, Amodo Design (2026). Original
21. DiLoCo: Distributed Low-Communication Training of Language Models, A. Douillard et al. (2024). Original
22. Does Distributed Training Undermine Compute Governance?, R. Rahman (2026). Original