P010 Chip identity and fingerprinting
Is there a clone-resistant chip or board identifier, or a fingerprint that tells apart individual chips of the same model, with a published false-match rate and cost that a prover cannot defeat by swapping in another chip?
Clone-resistant chip/board identifiers and same-SKU die fingerprints.
Terms: same-SKU die: an individual chip of the same product model as others (SKU = a product/stock code). PUF (physically unclonable function): a chip's unique, hard-to-copy fingerprint arising from tiny manufacturing variations. S-parameters: radio-frequency measurements of a board, usable as such a fingerprint.
Status
Started Software (atomic-race) fingerprint telling apart 24 same-model H200 GPUs rented on vast.ai, 480 runs; up to 100% paired re-identification. The authors' own limitations: not validated at larger scale; stability over time, transport and temperature not established; security against an adversary who emulates the GPU or runs the function faster not tested, so the headline's 'cannot defeat by swapping in another chip' clause and a published false-match rate at scale remain open. Lower-ranked evidence: Alpay & Alpay (2026) measure a per-SM latency map stable over a six-hour RTX 5090 run and 100% leave-one-out separation across distinct Blackwell dies, but leave same-model die separation to prior work; Safa, Mosavirik & Tajik (2023) distinguish an aged STM32 from 10–12 genuine ones by S-parameters with no error rates. No cost figure in any held source.
we evaluated the function across 24 Nvidia H200 GPUs, 2 seeds, and 10 runs each … fingerprinting function that achieved up to 100% accuracy in small-scale tests
Posed by
- Scher & Thiergart (2024), Mechanisms to Verify International Agreements About AI Development (MIRI):
Basic serial numbers are not adversarially robust.
(Building Blocks › Chip identity verification, p. 68, L3126–3127) (source states the problem but not the quantity the question asks for)
Also raised in: Reuel, Bucknall et al. (2025), Open Problems in Technical AI Governance, compute sections; Aarne, Fist & Withers (2024), Secure, Governable Chips (CNAS); Shavit (2023), What Does It Take to Catch a Chinchilla? Verifying Rules on Large-Scale Neural Network Training via Compute Monitoring; Abdelghafar & Kulp (2026), Privacy-Preserving AI Verification via Minimal Information Disclosure; Ansari (2026), Hardware-Level Governance of AI Compute: A Feasibility Taxonomy for Regulatory Compliance and Treaty Verification; Alpay & Alpay (2026), Unprivileged Topology Certificates for Cloud GPU Attestation, arXiv 2606.24934; AI Security Forum, Requests for Discussion.
Sub-questions
| ID | Sub-question | What the sources report |
|---|---|---|
| P010.1 | Is there a chip/board identifier (fuse, engraving, key or PUF) with demonstrated clone and modification resistance under invasive attack, a published false-accept rate, and stated cost — so a Prover cannot present a substitute with matching digital and physical serial? | partly answered |
| P010.2 | Do PUFs on a current advanced node reproduce reliably (bit error <= a stated bound) across temperature and ageing, and resist modelling attacks? | open |
| P010.3 | Does a per-SM latency-map die fingerprint separate dies of the same SKU at a stated accuracy, including multi-die parts (B200)? | a source suggests an answer (not checked) |
| P010.4 | Can an adversary emulate a target die's per-SM latency map on another die well enough to pass classification (defeating relay binding)? | a source suggests an answer (not checked) |
| P010.5 | Can S-parameter (RF scattering) analysis distinguish a genuine from a counterfeit or remarked data-centre GPU package? | partly answered |
| P010.6 | Can reported device properties used for identity (e.g. NVML memory capacity, H100 vs B200) be spoofed by driver modification without detection by current attestation? | open |
| P010.7 | Does any shipping accelerator support chip-to-chip mutual attestation over its scale-up interconnect, and has its spoof resistance been independently tested? | implied by a source, not stated as open |
Sub-questions not yet stated precisely enough to check
| ID | Sub-question |
|---|---|
| P010.8 | Can chip identity be established when vendor attestation is unavailable or the root of trust is compromised, given fingerprints need a cooperative host and do not prove firmware integrity? |
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 P010.1: Is there a chip/board identifier (fuse, engraving, key or PUF) with demonstrated clone and modification resistance under invasive attack, a published false-accept rate, and stated cost — so a Prover cannot present a substitute with matching digital and physical serial?
For P010.2: Do PUFs on a current advanced node reproduce reliably (bit error <= a stated bound) across temperature and ageing, and resist modelling attacks?
For P010.3: Does a per-SM latency-map die fingerprint separate dies of the same SKU at a stated accuracy, including multi-die parts (B200)?
For P010.5: Can S-parameter (RF scattering) analysis distinguish a genuine from a counterfeit or remarked data-centre GPU package?
* Full text not reviewed; more likely to contain errors.
Data
- problems.csv — one row per problem
- problem_status.csv — status, scope and evidence per problem
- problem_sources.csv — the posed-by quote and other documents per problem
- leaves.csv — all sub-questions, including those without a checkable form
- leaf_sources.csv — every source row behind each sub-question