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 and Timed challenge-response and memory-occupation challenges

No new hardware needed

Sampled inference recomputation and Timed challenge-response and memory-occupation challenges

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

• Side-channel suppression for isolated facilities, RAND secure inference data center (SIDC) design: Analysis
• Sampled inference recomputation: Analysis
• Timed challenge-response and memory-occupation challenges: Analysis

Family findings

• Side-channel suppression for isolated facilities

  Context for RAND secure inference data center (SIDC) design. These findings concern the family or other implementations; applicability must be checked against their stated scope.

  • Supply-chain implants may evade inspection in Side-channel suppression for isolated facilities

    Cankaya identifies malicious hardware embedded deep in purchased components as a residual risk that visual inspection and disassembly may not catch. He notes that radiographic examination under high-security standards could mitigate it. ²

    Theoretical argument · Significant · Open. On the record

  • Openings for airflow, power and optics weaken shielding in Side-channel suppression for isolated facilities

    Cankaya notes that keeping attenuation high while passing high-power airflow, cabling and optical links adds complexity beyond existing shielded-enclosure specifications. ²

    Theoretical argument · Significant · Open. On the record

  • Inspection assumptions may not hold in Side-channel suppression for isolated facilities

    The design's statistical argument assumes that visual or disassembly inspection catches every flaw that is present in a sampled unit. Cankaya is unsure whether destructive teardowns are defence-dominant or offence-dominant. ²

    Open question · Significant · Open. On the record

Open significant flaws

5 flaws in 3 mechanisms

• Everything rests on the trusted setup in RAND secure inference data center (SIDC) design

  Reference measurements for model weights and reference data are established in a trusted setup phase. The report states that the system cannot detect compromise that happened before ingestion if the trusted setup itself is compromised. ¹

  Theoretical argument · 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. ^{3 11 14}

  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. ^{3 15}

  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. ^{4 5}

  Theoretical argument · Significant · Open. On the record

• Remote memory narrows the timing margin in Timed challenge-response and memory-occupation challenges

  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. ¹⁶

  Theoretical argument · Significant · Open. On the record

  Related mechanism R2 Bandwidth limits and compartmentalization: Physical disconnection is proposed to exclude remote memory between the separated groups during a challenge. It depends on the isolation boundary being enforced. A pointer, not evidence that this flaw is mitigated. Add

Open minor flaws

3 mechanisms with minor findings

• Security weakens over long operation in RAND secure inference data center (SIDC) design

  The authors claim that the facility can withstand attacks at the OC5 level for a five-year operational period. They expect its ability to withstand long OC5 campaigns to become less robust the longer the facility remains in operation. ¹

  Theoretical argument · Minor · Open. On the record

• 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. ⁴

  Open question · Minor · Open. On the record

• Error rates not quantified in Timed challenge-response and memory-occupation challenges

  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. ¹⁸

  Open question · Minor · Open. On the record

Not yet demonstrated

R1 Side-channel suppression for isolated facilities

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 “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.
• R2 Bandwidth limits and compartmentalization Add
  • 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.
• R1 Network taps and certifiers Add
  • 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.
• R3 Model identity attestation Add
  • Side-channel suppression for isolated facilities depends on it.

Missing prerequisites

• R3 Model identity attestation needed by Side-channel suppression for isolated facilities Add

Blockers

12 blockers recorded

• Side-channel suppression for isolated facilities
  • No prototype exists; RAND recommends prototyping key security features and integration now. Adversarial validation ¹
  • The report describes internal integrity checks, audit logging and accreditation, but no way for a party outside the operator to verify the facility's properties. Access & governance ¹
  • Human review of every prompt and response makes each request take three to five minutes, with the review steps as the rate-limiting factor. Performance & compatibility ¹
  • Detailed design information is withheld from the public report and is to be evaluated privately with stakeholders, which limits independent public scrutiny. Access & governance 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 ^{10 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 ^{3 10}
  • 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 ^{10 11}
  • 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 ^{4 9}
  • 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 ^{4 15 16}
• Timed challenge-response and memory-occupation challenges
  • No network-level memory challenge across data-centre servers has been demonstrated. Adversarial validation ¹⁶
  • Challenges that fill memory displace workloads; filling a pod's volatile memory takes tens of minutes and SSDs take hours. Performance & compatibility ^{16 18}
  • Outside help, such as remote memory, must be excluded during challenges. Coverage & hidden compute. Waits on Bandwidth limits and compartmentalization ¹⁶

From the family or selected implementation's record. ?

Model weights

Depends on the design for Sampled inference recomputation.

Not involved: Timed challenge-response and memory-occupation challenges.

Unspecified for Side-channel suppression for isolated facilities. Check the implementation record.

Inputs and outputs

Depends on the design for Sampled inference recomputation.

Not involved: Timed challenge-response and memory-occupation challenges.

Unspecified for Side-channel suppression for isolated facilities. Check the implementation record.

Training data

Not involved: Sampled inference recomputation and Timed challenge-response and memory-occupation challenges.

Unspecified for Side-channel suppression for isolated facilities. Check the implementation record.

Side-channel suppression for isolated facilities

• R1 AI 2040 inference-only verification stack Proposed architecture, AI Futures Project
• R1 Low-trust AI compute verification system overview Proposed architecture, Machine Intelligence Research Institute
• R1 RAND secure inference data center (SIDC) design Proposed architecture, RAND

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

Timed challenge-response and memory-occupation challenges

• R1 Data-centre memory challenging Proposed architecture, Machine Intelligence Research Institute
• R2 GPU contention probes Research prototype
• R1 Low-trust AI compute verification system overview Proposed architecture, Machine Intelligence Research Institute
• R2 SAGE Research prototype
• R2 VRAM-residency challenge Research prototype

1. Highly Secure Inference Data Centers: A Vertically Integrated Strategy for Security Engineering, S. F. Comer et al. (2026). Original
2. Suppressing Side Channels in an Untrusted Data Center via Retrofitted Defenses, N. Cankaya (2026). Original
3. Verifying LLM Inference to Detect Model Weight Exfiltration, R. Rinberg et al. (2025). Original
4. DiFR: Inference Verification Despite Nondeterminism, A. Karvonen et al. (2025). Original
5. TOPLOC: A Locality Sensitive Hashing Scheme for Trustless Verifiable Inference, J. M. Ong et al. (2025). Original
6. PrimeIntellect-ai/toploc (GitHub repository), Prime Intellect (2025). Original
7. INTELLECT-2: A Reasoning Model Trained Through Globally Decentralized Reinforcement Learning, Prime Intellect Team et al. (2025). Original
8. adamkarvonen/difr (GitHub repository), A. Karvonen (2025). Original
9. Scaling Recomputation Inference Verification, Amodo Design (2026). Original
10. AI 2040 Plan A — Verification SITREP, Amodo Design (2026). Original
11. Adversarial Entropy Inflation Against Gumbel-Based Inference Verification, N. Kezins (2026). Original
12. SYNTHETIC-2 Release: Four Million Collaboratively Generated Reasoning Traces, Prime Intellect (2025). Original
13. An Inference Verification Prototype — Stage 1, Amodo Design (2026). Original
14. Bit-Exact AI Inference Verification Without Performance Tradeoffs, N. Cankaya (2026). Original
15. Example Schemes for Verifying High-Stakes AI Agreements, Amodo Design (2026). Original
16. A System Overview for Near-Term, Low-Trust AI Compute Verification, N. Cankaya (2026). Original
17. Verification Plan, R. Dean (2026). Original
18. Timing and Memory Telemetry on GPUs for AI Governance, S. K. Monfared et al. (2026). Original
19. SAGE: Software-based Attestation for GPU Execution, A. Ivanov et al. (2023). Original
20. SWATT: SoftWare-based ATTestation for Embedded Devices, A. Seshadri et al. (2004). Original
21. Proofs of Space, S. Dziembowski et al. (2015). Original
22. Secure Code Update for Embedded Devices via Proofs of Secure Erasure, D. Perito & G. Tsudik (2010). Original
23. Software-Based Memory Erasure with Relaxed Isolation Requirements, S. Bursuc et al. (2024). Original
24. On the Difficulty of Software-Based Attestation of Embedded Devices, C. Castelluccia et al. (2009). Original
25. Refutation of "On the Difficulty of Software-Based Attestation of Embedded Devices", A. Perrig & L. van Doorn (2010). Original