P011 Matching fabricated silicon to the approved design

Can a party other than designer and fab confirm fabricated silicon matches the approved design and detect a deliberately inserted mismatch at a stated rate?

Checking fabricated silicon against the approved design.

Status

Analyzed A proposal only: the non-manufacturing side samples fabricated components and destructively reverse-engineers them against the manufacturer's design documents; flexHEG Part 2 adds open-sourcing the guarantee-processor design for comparison with physical scans of randomly selected chips, and Ilhan et al. (2026) analyse physical inspection and sampling as the direct check. No held document reports a third party checking a fabricated die against its design, and none gives a detection rate on a deliberately inserted mismatch; the only rate is Agrawal et al. (2007), a circuit-level power simulation (49/49 simulated Trojan chips caught, 1/49 false alarm) on a 28k-gate RSA core, no silicon measured. Puschner et al.'s real-world SEM Trojan-detection case study is named by the Proofworks neutral-compute note but not held.

some of the existing components could be sampled for rigorous, destructive reverse-engineering to check that they match the design documents provided by the manufacturer.
Petrie, Aarne, Ammann & Dalrymple (2025), Overview of Flexible Hardware-Enabled Guarantees (flexHEG Part 1) (2025-04), L538–540

Posed by

Also raised in: Baker, Kulp, Marks, Brundage & Heim (2025), Verifying International Agreements on AI: Six Layers of Verification for Rules on Large-Scale AI Development and Deployment (RAND WR-A4077-1); Scher & Thiergart (2024), Mechanisms to Verify International Agreements About AI Development (MIRI); Petrie, Aarne, Ammann & Dalrymple (2025), Overview of Flexible Hardware-Enabled Guarantees (flexHEG Part 1); Petrie & Aarne (2025), Technical Options for Flexible Hardware-Enabled Guarantees (flexHEG Part 2); Ilhan, Withers, Gietz & Harack (2026), Verifiable Semiconductor Manufacturing (Oxford AIGI); Ansari (2026), Hardware-Level Governance of AI Compute: A Feasibility Taxonomy for Regulatory Compliance and Treaty Verification; Amodo (2026), AI 2040 Plan A — How You Can Help; Proofworks, Living Verification Documents.

Sub-questions

IDSub-questionWhat the sources report
P011.1Has any method checked a scanned leading-edge die against its HDL/design at gate level by a party other than designer and fab, detecting a deliberately inserted mismatch at a stated rate and analyst-hours?open
P011.2What is the most advanced chip cooperatively fabricated from an open design (trailing node) and fully verified by imaging, and is it sufficient for a guarantee processor or tap, given open GDS is impossible at the leading edge?partly answered
P011.3Does sampled (destructive) testing of fabricated chips detect an omitted, altered or backdoored governance block at a stated sample size and detection rate?implied by a source, not stated as open
P011.4Can a non-destructive test distinguish an approved accelerator die from one with a small added implant, or a backdoored chip from a clean one of the same SKU (in-field), with stated TPR/FPR?open
P011.5Can a zero-knowledge or managed-access protocol prove a mask set / netlist matches a reviewed design (netlist-to-layout equivalence) without revealing it, at non-trivial size?open
P011.6Can a third party check post-OPC / dummy-fill masks against the designer's GDSII and confirm no functional change at an advanced node?open
P011.7Can an independent equivalence check confirm synthesised netlist and masks match RTL for accelerator-scale designs (against compromised EDA tools)?open
P011.8Does any layout-integrity check detect published ECO-based trojan insertions by a rogue foundry insider before tape-out?open

Possibly relevant work found by automated search

Found by AI agents searching for this problem's sub-questions, mostly by keyword and citation in the OpenAlex scholarly database and on the web; a second AI pass reviewed each one, and we spot-checked some by hand. They seemed reasonably related, as far as one can tell from outside each paper's field. Some come from neighbouring fields, such as arms control and financial audit, and bear on the problem by analogy. Not reviewed against the status above; a paper listed here does not change that status. 928 of 1052 quotes checked word for word against the paper's text.

For P011.1: Has any method checked a scanned leading-edge die against its HDL/design at gate level by a party other than designer and fab, detecting a deliberately inserted mismatch at a stated rate and analyst-hours?

For P011.2: What is the most advanced chip cooperatively fabricated from an open design (trailing node) and fully verified by imaging, and is it sufficient for a guarantee processor or tap, given open GDS is impossible at the leading edge?

For P011.3: Does sampled (destructive) testing of fabricated chips detect an omitted, altered or backdoored governance block at a stated sample size and detection rate?

For P011.4: Can a non-destructive test distinguish an approved accelerator die from one with a small added implant, or a backdoored chip from a clean one of the same SKU (in-field), with stated TPR/FPR?

For P011.5: Can a zero-knowledge or managed-access protocol prove a mask set / netlist matches a reviewed design (netlist-to-layout equivalence) without revealing it, at non-trivial size?

For P011.8: Does any layout-integrity check detect published ECO-based trojan insertions by a rogue foundry insider before tape-out?

Data