Proposal Explorer

Add claims to see which ones the mechanisms address.

Operational use

• Operational use Deterministic and bit-exact inference for reproducing open-model inference from receipts in Gensyn's information-market service

Built for an adversarial prover

Deterministic and bit-exact inference and Bandwidth limits and compartmentalization

No new hardware needed

Deterministic and bit-exact inference

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.

• Deterministic and bit-exact inference: Analysis

Family findings

• Bandwidth limits and compartmentalization

  Context for AI 2040 inference-only verification stack. Findings from the mechanism family appear here as context. They apply to an implementation only when its own record lists them, under the conditions stated there.

  • 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. ^{19 20 21}

    Open question · Theoretical argument. 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. ²¹

    Known failure · 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. ²¹

    Scope limitation · Theoretical argument. 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. ²¹

    Scope limitation · Open question. 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. ²²

    Known failure · Theoretical argument · Significant · Open. On the record

Open significant failures

1 failure in 1 mechanism

• A recomputation family degrades against prompt-controlling adversaries in AI 2040 inference-only verification stack

  The plan's companion page names DiFR among the recomputation schemes being tested. An independent study found that Gumbel-based inference verification, the family that includes Token-DiFR, leaks roughly twice as many bits per token when the adversary chooses prompts that disrupt the text's structure, across six models of 1 to 32 billion parameters. The slowdown it imposes on a hidden-signalling adversary falls from 146–254 times under benign prompts to 60–118 times. The attack weakens the bound on hidden information in outputs, which the plan relies on to keep undeclared results from leaving. ^{12 13 17 18}

  Known failure · Demonstrated attack · Significant · Open. On the record

Scope limitations

• Some kernels remain genuinely nondeterministic in Deterministic and bit-exact inference

  The bit-exact work separates kernels that are deterministic but not batch-invariant from truly nondeterministic ones that use atomic functions. Some integer de-quantization kernels use atomic additions and remain nondeterministic, so exact replay needs backends that avoid them. ¹

  Scope limitation · Open question. On the record

• Cross-hardware replay relies on reverse-engineered, closed behaviour in Deterministic and bit-exact inference

  Emulating one GPU's rounding on another requires reverse-engineering tensor-core arithmetic and modelling proprietary kernel choices. Hawkeye covers a subset of NVIDIA architectures and states that attention and other higher-level operations need further reverse engineering. For the bit-exact emulator, a proprietary Hopper kernel family is an open edge case. ^{1 3}

  Scope limitation · Open question. On the record

• The recomputation server must be trusted in AI 2040 inference-only verification stack

  The plan calls the integrity of the recomputation server an extremely important aspect, and its argument that sampling verifies all outputs assumes that the server's computations and outputs can be trusted. The companion page notes that the server sits inside the prover's facility, possibly under the prover's physical control, and that hardening it against integrity attacks needs significant research. Amodo rates recomputation-server security as not on track. ^{12 13 15}

  Scope limitation · Theoretical argument. On the record

• Spare compute is not verified in AI 2040 inference-only verification stack

  The plan states that it does not verify that spare compute is unused for unapproved workloads, because this seems very challenging. It relies instead on side-channel bounds and memory wipes, so that the only results that persist are verified inference outputs. ¹²

  Scope limitation · Theoretical argument. On the record

Not yet demonstrated

Proposed Bandwidth limits and compartmentalization

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

• Research demo Tamper evidence for verifier devices Add
  • Bandwidth limits and compartmentalization waits on it. Checking that taps are correctly installed and stay in place at scale is not a solved problem, and hardening the recomputation server inside the prover's facility needs significant research.
• Proposed Memory wiping and proofs of secure erasure Add
  • Bandwidth limits and compartmentalization waits on it. Memory wiping may use existing algorithms, but hardware testing is at an early stage.
• Proposed Network taps and certifiers Add
  • Bandwidth limits and compartmentalization waits on it. Passive optical taps work at 400G, but the 800G and 1600G line rates now arriving in data centres are undemonstrated.
• Proposed Side-channel suppression for isolated facilities Add
  • Bandwidth limits and compartmentalization waits on it. There is no plan yet for quickly scaling side-channel defences on a frontier cluster; only early theoretical pieces exist.
• Proposed Whole-workload recomputation (reproducible packets) Add
  • Bandwidth limits and compartmentalization waits on it. A fully reproducible inference stack needs substantial software and tooling, and per-packet network reproducibility may need considerable software, firmware and possibly hardware work.
• Operational use Sampled inference recomputation Add
  • Bandwidth limits and compartmentalization depends on it.

Missing prerequisites

• Operational use Sampled inference recomputation needed by Bandwidth limits and compartmentalization Add
• Proposed Whole-workload recomputation (reproducible packets) needed by Bandwidth limits and compartmentalization Add

Blockers

10 blockers recorded

• Deterministic and bit-exact inference
  • Batch-invariant kernels cost throughput: in Thinking Machines' Qwen3-8B test, an improved deterministic build took 42 s against 26 s for vLLM's default, and SGLang reports an average 34.35% slowdown on its FlashInfer and FlashAttention 3 backends. Performance & compatibility ^{2 4}
  • Coverage is incomplete: the bit-exact emulator targets dense blocks on NVIDIA GPUs and excludes mixture-of-experts inference and training, and vLLM's batch-invariant mode is in beta, with open work on AMD hardware and speculative decoding. Performance & compatibility ^{1 5 23}
  • Amodo's status page for the AI 2040 verification plan rates a reproducible inference stack for that plan as 'not started'. Performance & compatibility ¹⁵
  • Exact replay requires the prover to disclose weights, software versions, parallelism and batch sizes to whoever recomputes. Privacy & leakage ^{1 11}
• Bandwidth limits and compartmentalization
  • A fully reproducible inference stack needs substantial software and tooling, and per-packet network reproducibility may need considerable software, firmware and possibly hardware work. Performance & compatibility. Waits on Whole-workload recomputation (reproducible packets) ¹³
  • Passive optical taps work at 400G, but the 800G and 1600G line rates now arriving in data centres are undemonstrated. Performance & compatibility. Waits on Network taps and certifiers ¹³
  • Checking that taps are correctly installed and stay in place at scale is not a solved problem, and hardening the recomputation server inside the prover's facility needs significant research. Hardware trust. Waits on Tamper evidence for verifier devices ^{13 15}
  • There is no plan yet for quickly scaling side-channel defences on a frontier cluster; only early theoretical pieces exist. Coverage & hidden compute. Waits on Side-channel suppression for isolated facilities ¹³
  • Memory wiping may use existing algorithms, but hardware testing is at an early stage. Coverage & hidden compute. Waits on Memory wiping and proofs of secure erasure ¹³
  • Robust red-teaming of recomputation schemes has not started, and most algorithm development remains academic. Adversarial validation ^{13 15}

From the family or selected implementation's record. ?

Model weights

Depends on the design for Deterministic and bit-exact inference.

Unspecified for Bandwidth limits and compartmentalization. Check the implementation record.

Inputs and outputs

Depends on the design for Deterministic and bit-exact inference.

Unspecified for Bandwidth limits and compartmentalization. Check the implementation record.

Training data

Not involved: Deterministic and bit-exact inference.

Unspecified for Bandwidth limits and compartmentalization. Check the implementation record.

Deterministic and bit-exact inference

• Research demo Batch-invariant inference kernels (Thinking Machines) Open-source project, Thinking Machines Lab
• Operational use Verde and RepOps (Gensyn) Product, Gensyn
• Proposed Low-trust AI compute verification system overview Proposed architecture, Machine Intelligence Research Institute

Bandwidth limits and compartmentalization

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

1. Bit-Exact AI Inference Verification Without Performance Tradeoffs, N. Cankaya (2026). Original
2. Defeating Nondeterminism in LLM Inference, H. He & Thinking Machines Lab (2025). Original
3. Hawkeye: Reproducing GPU-Level Non-Determinism, E. Badash et al. (2026). Original
4. Towards Deterministic Inference in SGLang and Reproducible RL Training, The SGLang Team (2025). Original
5. Batch Invariance (vLLM documentation), vLLM project (2026). Original
6. gensyn-ai/ree: Gensyn Reproducible Execution Environment (GitHub repository), Gensyn (2026). Original
7. EigenCloud Brings Verifiable AI to Mass Market with EigenAI and EigenCompute Launches, EigenCloud (2025). Original
8. Building Delphi: Pricing, Settlement, and Agentic Trading, D. Jedamski (2026). Original
9. Reproducible Execution Environment (REE) (Gensyn documentation), Gensyn (2026). Original
10. What is Delphi? (Delphi documentation), Gensyn (2026). Original
11. A System Overview for Near-Term, Low-Trust AI Compute Verification, N. Cankaya (2026). Original
12. Verification Plan, R. Dean (2026). Original
13. Get Involved in Verification, AI Futures Project (2026). Original
14. Verifying international AI deals: Plan A, the state-of-play, and what you can do to help, T. Milton et al. (2026). Original
15. AI 2040 Plan A — Verification SITREP, Amodo Design (2026). Original
16. Fitting a Network TAP to our Inference Verification Prototype, Amodo Design (2026). Original
17. Verifying LLM Inference to Detect Model Weight Exfiltration, R. Rinberg et al. (2025). Original
18. Adversarial Entropy Inflation Against Gumbel-Based Inference Verification, N. Kezins (2026). Original
19. DiLoCo: Distributed Low-Communication Training of Language Models, A. Douillard et al. (2024). Original
20. Does Distributed Training Undermine Compute Governance?, R. Rahman (2026). Original
21. Traffic Shaping for Workload Classification, Lucid Computing (2026). Original
22. The Tray as a Bandwidth Boundary, Amodo Design (2026). Original
23. [Feature]: Batch Invariant Feature and Performance Optimization (vLLM issue #27433), vLLM project contributors (2025). Original