P026 Location verification of chips
What location precision and false-accept rate can network-delay geolocation of accelerators achieve against an adversary who adds delay, relays, or spoofs?
Delay-based and network-measurement geolocation and co-location of accelerators against adversarial delay, relays and spoofing.
Terms: N: the number of chips taking part in a co-location challenge.
Status
Started The authors' own measurement of their own method on rented cloud GPUs with cooperative hosts: network-delay geolocation against public RIPE Atlas landmarks places a B200 within 44 km and a V100 within 270 km (79-298 km depending on landmark density) and rejects 11 of 11 decoy sites; the decoys are other locations, not an adversary adding delay or relaying, and the paper says a high round-trip time is no evidence of absence from a forbidden region and the scheme is not tamper-proof, so the headline's false-accept rate against added delay, relays or spoofing is not reported. Timed co-location challenges across N chips: no work. Robustness of on-chip GPS to commercial spoofers is a related sub-question, not in the headline, with no work.
169 RIPE Atlas probes place the B200 within 44 km of its claimed datacentre and reject all 11 decoy sites.
Posed by
- Reuel, Bucknall et al. (2025), Open Problems in Technical AI Governance, compute sections:
Accurately detecting a chip’s location remains an open challenge.
(§5.2.1, p.28, L1494 (Q49)) (source states the problem but not the quantity the question asks for)
Also raised in: Harack et al. (2025), Verification for International AI Governance (Oxford AIGI); 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); Aarne & Petrie (2025), International Security Applications of Flexible Hardware-Enabled Guarantees (flexHEG Part 3); Aarne, Fist & Withers (2024), Secure, Governable Chips (CNAS); Ansari (2026), Hardware-Level Governance of AI Compute: A Feasibility Taxonomy for Regulatory Compliance and Treaty Verification; Rahman (2026), Does Distributed Training Undermine Compute Governance? (MIRI); Wasil, Reed, Miller & Barnett (2024), Verification Methods for International AI Agreements; Alpay & Alpay (2026), Unprivileged Topology Certificates for Cloud GPU Attestation, arXiv 2606.24934; AI Security Forum, Requests for Discussion; Dean, Verification Plan (AI 2040).
Sub-questions
| ID | Sub-question | What the sources report |
|---|---|---|
| P026.1 | What location precision (km) and confidence does latency-based geolocation against trusted landmarks achieve against an adversary who adds delay or relays? | partly answered |
| P026.2 | How many landmark servers (and how close: <100 km near borders) are needed to verify country-level location for a region, with what false-accept rate? | partly answered |
| P026.3 | Has an on-chip location mechanism rejected a commercially available GPS spoofer in a test, with a reported false-accept rate? | open |
| P026.4 | Can a coarse-location scheme prove 'inside a given country' without revealing position to better than a stated radius? | partly answered |
| P026.5 | Has a timed co-location challenge been demonstrated that N chips answer only if they share one site's interconnect, with a false-accept rate when split across sites? | open |
| P026.6 | What fraction of accelerators must carry location attestation for an evader using only unverified chips (or distributed training) to need a stated multiple of the hardware cost? | a source suggests an answer (not checked) |
Sub-questions not yet stated precisely enough to check
| ID | Sub-question |
|---|---|
| P026.7 | Since added delay can only make a chip look farther away, how should a scheme treat imprecise or delay-inflated results (e.g. licence revocation) without excessive false alarms? |
| P026.8 | Can location tracking detect covert transfers, faked on-sales or faked chip deaths? |
| P026.9 | What is the threat model for network-based location verification (spoofed GPS, compromised network infrastructure)? |
| P026.10 | Can monitoring data streams be protected against location triangulation by sophisticated adversaries? |
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 P026.1: What location precision (km) and confidence does latency-based geolocation against trusted landmarks achieve against an adversary who adds delay or relays?
For P026.3: Has an on-chip location mechanism rejected a commercially available GPS spoofer in a test, with a reported false-accept rate?
For P026.4: Can a coarse-location scheme prove 'inside a given country' without revealing position to better than a stated radius?
* 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