Proposal Explorer
ResetA verification proposal from the AI Verification Tech Map, from its records of 2026-10-09. https://trustbutveri.fyi/explorer/?mechanisms=M-0022,M-0014&implementations=M-0014:I-0010&ready=R4
Analysis
Applied filters: Legacy independent-evaluation filter: R4 (legacy). Not set (Any): Prover, Verifier devices on site, Prover cooperation, Chips, Attack testing, Keep hidden from the verifier.
Overview
| Mechanism | Development | Security evidence | Open failures |
|---|---|---|---|
| Side-channel suppression for isolated facilities ⚠ excluded by your filters: legacy evaluation filter | Proposed | Published security analysis | none |
| Bandwidth limits and compartmentalizationRAND secure inference data center (SIDC) design ⚠ excluded by your filters: legacy evaluation filter | Proposed | Published security analysis | noneFamily context below |
- Open failures: n critical n significant n minor
- ⚠ Dimmed: excluded by your filters, with the conflicting field highlighted
Claim coverageNo claims yet
Add claims to see which ones the mechanisms address.
PropertiesNone recorded
- Not counted
- Excluded by your filters: Side-channel suppression for isolated facilities and Bandwidth limits and compartmentalization
Attack testing2 mechanisms with published testing
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.
- Side-channel suppression for isolated facilities: Analysis · excluded by filters
- Bandwidth limits and compartmentalization, RAND secure inference data center (SIDC) design: Analysis · excluded by filters
Limits1 family with findings to check · 3 scope limitations · 2 open questions · 2 excluded by filters · 2 not yet demonstrated
- Excluded by your filters
- Side-channel suppression for isolated facilities Legacy independent-evaluation filter (R4): no matching legacy code.
- Bandwidth limits and compartmentalization Legacy independent-evaluation filter (R4): no matching legacy code.
- Family findings
- Bandwidth limits and compartmentalization
Context for RAND secure inference data center (SIDC) design. 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. 3 4 5
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. 5
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. 5
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. 5
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. 6
Known failure · Theoretical argument · Significant · Open. On the record
- Bandwidth limits and compartmentalization
- Scope limitations
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. 1
Scope limitation · Theoretical argument. On the record
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. 2
Scope limitation · Theoretical argument. On the record
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. 2
Scope limitation · Theoretical argument. On the record
- Open questions
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. 1
Open question · Theoretical argument. 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. 1
Open question · Open question. On the record
- Not yet demonstrated
- Proposed Side-channel suppression for isolated facilities and Proposed Bandwidth limits and compartmentalization
Possible additionsNone found on the map
Mechanisms on the map that are not in the proposal. Pointers, not recommendations.
The records connect no other mechanism to this proposal's gaps, open failures or dependencies.
Dependencies6 blockers
- Blockers
6 blockers recorded
- Side-channel suppression for isolated facilities
- Bandwidth limits and compartmentalization
- No prototype exists; RAND recommends prototyping key security features and integration now. Adversarial validation 2
- 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 2
- 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 2
- Detailed design information is withheld from the public report and is to be evaluated privately with stakeholders, which limits independent public scrutiny. Access & governance 2
What the verifier sees1 unspecified
From the family or selected implementation's record.
- Model weights
Not involved: Side-channel suppression for isolated facilities.
Unspecified for Bandwidth limits and compartmentalization. Check the implementation record.
- Inputs and outputs
Not involved: Side-channel suppression for isolated facilities.
Unspecified for Bandwidth limits and compartmentalization. Check the implementation record.
- Training data
Not involved: Side-channel suppression for isolated facilities.
Unspecified for Bandwidth limits and compartmentalization. Check the implementation record.
Exposure notes
- Side-channel suppression for isolated facilities: Shields and filters a facility; it does not handle model data.
- Bandwidth limits and compartmentalization, RAND secure inference data center (SIDC) design: This Explorer has no asset-specific exposure assessment for this implementation. Check its source and deployment assumptions.
Implementations3 systems
- Side-channel suppression for isolated facilities
- Proposed AI 2040 inference-only verification stack Proposed architecture, AI Futures Project
- Proposed Low-trust AI compute verification system overview Proposed architecture, Machine Intelligence Research Institute
- Proposed RAND secure inference data center (SIDC) design Proposed architecture, RAND
- 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
Sources6 cited
- Suppressing Side Channels in an Untrusted Data Center via Retrofitted Defenses, N. Cankaya (2026). Original
- Highly Secure Inference Data Centers: A Vertically Integrated Strategy for Security Engineering, S. F. Comer et al. (2026). Original
- DiLoCo: Distributed Low-Communication Training of Language Models, A. Douillard et al. (2024). Original
- Does Distributed Training Undermine Compute Governance?, R. Rahman (2026). Original
- Traffic Shaping for Workload Classification, Lucid Computing (2026). Original
- The Tray as a Bandwidth Boundary, Amodo Design (2026). Original
Share the link to this proposal. This proposal is also available as plain text and JSON.
Filter mechanisms
Filters apply to mechanisms only. They describe the setting a proposal is for, and all are off by default. A mechanism that a filter rules out is flagged and does not count towards claim coverage. Selected implementations use their own record fields. A match means not excluded; conditional or unspecified exposure stays with a note. Passing a filter does not establish that the assumptions hold in a deployment.
0 of 25 mechanisms match · Clear all
Prover
How far can the party being checked be trusted?
The prover is the party being checked. Semi-trusted designs rely on part of its stack: usually the chip vendor's hardware root of trust, its firmware or counters, or its supply-chain records. Adversarial designs aim to hold even if it cheats wherever the checks allow, within their stated assumptions.
Keeps mechanisms whose threat model holds against at least this prover. Adversarial is the strongest assumption. Definitions
Verifier devices on site
May the verifier install its own hardware at the prover's sites?
Some mechanisms need a device the verifier owns or trusts at the prover's facility, such as a network tap, a bandwidth limiter or a sealed sensor. Choose Not allowed when the setting rules that out. Inspectors are not covered.
"Not allowed" removes mechanisms that need a retrofit device, such as a network tap or a sealed sensor. Definitions
Prover cooperation
How much must the prover take part?
Required: the prover takes part, for example by logging requests, producing proofs or opening records. Partial: some access, such as installing a device. Not required: works from outside, such as satellite imagery.
"Partial at most" removes mechanisms that need the prover's active participation. "Not required" keeps only those that work without it. Definitions
Chips
May the proposal depend on new chip designs?
New chip features take years to reach a deployed fleet and cover only chips made after they ship. Mechanisms that use shipping features, such as trusted execution environments or performance counters, stay.
"Existing chips only" removes mechanisms that need changes to future chip designs. Definitions
Legacy independent-evaluation filter
Development status
- Any25 match
- Proposed25 match
- Research demonstration14 match
- Operational use4 match
- Legacy independent-evaluation filter (selected)0 match
A level describes the public evidence for a mechanism's stated use, not its cost or feasibility. R3 can still have open critical flaws.
Exact legacy code: R4. Definitions
Attack testing
How hard has each mechanism been attacked in public?
The strongest published attempt to break the mechanism for its verification use: a security analysis, red-teaming by its developers or collaborators, or a red team independent of them.
Keeps mechanisms whose strongest published attack testing is at least this. Definitions
Keep hidden from the verifier
What must the verifier never see? Choose any.
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.
Removes mechanisms that show the asset to the verifier. Conditional or unspecified exposure stays with a note and needs checking against the privacy requirement.
Claims
A claim is something one party wants to verify about another party's AI hardware or software. Each claim's number shows how the proposal addresses it.
- Addressed. A mechanism in the proposal is aimed at this claim and is not excluded by the filters.
- Partly addressed. Only supporting mechanisms, or mechanisms aimed at it that the filters exclude.
- Unaddressed. No mechanism in the proposal addresses this claim.
Addressed means a mechanism in the proposal is aimed at the claim and is not excluded by your filters. It does not mean the claim is verified: check its assessed use, development status, security evidence, assumptions and findings.
Mechanisms
A mechanism is a general technique for verifying claims. Its badge is its development status for its stated use. An optional implementation choice narrows its assumptions, assessed use and claim links to that record. Lines join it to the claims it addresses. Click a line for details.Under its name it lists the claims it is aimed at or supports.
- Aimed at the claim: verifying it is a direct purpose of the mechanism.
- Supports the claim: helps verify it without being aimed at it.
- Faint: excluded by your filters, so it does not count towards claim coverage.
Overview
One row per mechanism in the proposal. Every mark comes from that mechanism's record, as listed in the panels below; what the verifier sees is the editors' reading of the record's text. Failure counts are per mechanism. Summary counts name mechanisms with open failures, not a sum of attacks. Choosing an implementation narrows each row to that record's assessed use; family findings remain as context. Findings are grouped as known failures, scope limitations and open questions. Only known failures count as failures. Counts are an inventory of published findings, not a risk score.
What the verifier sees
For model weights, inputs and outputs, and training data. This is the editors' reading of each mechanism's record (its threat model, how it works and its limitations), not a field of the record. Shown: the verifier sees it. Depends: on the design or variant, or the verifier sees only samples. Hidden: the verifier sees only commitments, hashes, proofs or results. Not involved: the record does not handle it. Unspecified: the selected implementation has no asset-specific assessment here.
Possible additions
Mechanisms on the map, not in the proposal, that the records connect to an unaddressed or partly addressed claim, an open failure or a dependency. They are pointers, not recommendations: each brings its own readiness level and findings, and none is claimed to close a failure. Links from failures are the editors' reading of the two records.
Start from a goal
A goal is something a rule or agreement about AI sets out to achieve. Choosing one loads the claims it needs verified. Your mechanisms and filters stay as they are.
- Cap frontier training2 direct, 3 supportingKeep every AI training run below an agreed amount of compute.
- Pause frontier AI development2 direct, 3 supportingStop new AI training runs and experiments for an agreed period, while existing models stay in service.
- Deploy only evaluated models2 direct, 3 supportingDeploy powerful AI models widely only after their risks have been evaluated and judged manageable.
- Prevent catastrophic misuse2 direct, 2 supportingKeep capable AI models from helping anyone carry out catastrophic attacks, such as biological or chemical ones.
- Prevent weight theft1 direct, 1 supportingKeep the weights of capable AI models from being copied out of the facilities that hold them.
- Enforce chip export controls1 directKeep export-controlled AI chips at the destinations they were authorised for.
Start from a published design
Choosing a design loads the mechanisms its record realises or depends on. If the proposal has no claims yet, it also loads the claims that record says the design addresses.
- AI 2040 inference-only verification stack7 mechanismsProposed architecture, AI Futures Project
- Low-trust AI compute verification system overview7 mechanismsProposed architecture, Machine Intelligence Research Institute
- RAND secure inference data center (SIDC) design2 mechanismsProposed architecture, RAND