P005 Facility air-gap and enclosure containment
Has an air-gapped facility or pod been red-teamed against electromagnetic, acoustic, optical and personnel channels, with a verified limit on what can leak out?
Air-gap and shielded-enclosure containment of facilities and pods.
Terms: air-gap: physically isolating a facility from outside networks. dB attenuation: how much a shield cuts a signal, in decibels (every 10 dB is a tenfold reduction). SCIF: a shielded, access-controlled secure room.
Status
Analyzed Desk study only: a literature survey of physical side channels and an example defence architecture for rack or pod enclosures, with an estimated bill of materials below 1% of cluster cost while tolerating single-digit kbit/s of covert bandwidth; the author calls it a two-week sprint, 'far from conclusive'. Nothing built; no facility or pod red-teamed; no verified leakage limit; the RF attenuation of cheaply welded sheet-metal enclosures is an assumption, not a measurement. Single-computer exfiltration demonstrations held in the collection (Fansmitter acoustic exfiltration at 15 bits/min, Deep-TEMPEST HDMI, Synesthesia screen acoustics) are attack measurements at PC scale, not containment tests, and are cross-references because the headline names facility or pod scale. The Proofworks workshop 'seeming consensus' that 1-bit air-gapping is feasible is an opinion.
I present an example of a defense architecture towards the end of this post. It leverages a synergy of cheap enclosures, sensors, jamming, unannounced hardware inspections and physical security
Posed by
- Harack et al. (2025), Verification for International AI Governance (Oxford AIGI):
it is unclear how an air gapped facility can be reliably verified
(§2.5.2.4 fn 203, p.63, L3446–3447)
Also raised in: Aarne & Petrie (2025), International Security Applications of Flexible Hardware-Enabled Guarantees (flexHEG Part 3); Cankaya (2026), Catching Misreporting About ML Hardware Use by Turning Noise into Signal, Part II; Cankaya (2026), Research Note: The Fundamentals and Feasibility of Secure Network Taps for Verifying AI Datacenter Use; Cankaya (2026), Suppressing Side Channels in an Untrusted Data Center via Retrofitted Defenses (MIRI); Proofworks, Living Verification Documents.
Sub-questions
| ID | Sub-question | What the sources report |
|---|---|---|
| P005.1 | Has an air-gapped facility been red-teamed against EM, acoustic, optical and personnel channels with a verified egress bound (e.g. <=1 bit per job)? | open |
| P005.2 | Has a prototype pod or replay enclosure been independently red-teamed with residual covert bandwidth measured (bps)? | open |
| P005.3 | What is the smallest RF/optical/acoustic exfiltration device a data-hall sweep reliably detects, with a reported miss rate? | open |
| P005.4 | Does a two-layer (one layer per party) SCIF survive a red-team covert-channel test that a single operator-built SCIF fails? | open |
| P005.5 | What attenuation (dB) do honeycomb-mesh vents achieve at rack-scale airflow? | open |
| P005.6 | Can a COTS SCIF pass a stated liquid-cooling load while keeping its shielding rating? | open |
| P005.7 | What attenuation (dB, 100 MHz-10 GHz) does a cheaply welded 1 mm steel pod enclosure with double-right-angle conduits achieve (target >= 62 dB)? | implied by a source, not stated as open |
| P005.8 | At what separation distance does EM/acoustic side-channel leakage between adjacent racks fall below a stated threshold? | implied by a source, not stated as open |
| P005.9 | What is the cost ratio of per-rack vs per-pod vs per-SuperPOD perimeters? | open |
Sub-questions not yet stated precisely enough to check
| ID | Sub-question |
|---|---|
| P005.10 | Is the enumerated list of independent physical side channels out of an enclosure complete? |
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 P005.1: Has an air-gapped facility been red-teamed against EM, acoustic, optical and personnel channels with a verified egress bound (e.g. <=1 bit per job)?
For P005.2: Has a prototype pod or replay enclosure been independently red-teamed with residual covert bandwidth measured (bps)?
For P005.3: What is the smallest RF/optical/acoustic exfiltration device a data-hall sweep reliably detects, with a reported miss rate?
For P005.4: Does a two-layer (one layer per party) SCIF survive a red-team covert-channel test that a single operator-built SCIF fails?
For P005.5: What attenuation (dB) do honeycomb-mesh vents achieve at rack-scale airflow?
For P005.6: Can a COTS SCIF pass a stated liquid-cooling load while keeping its shielding rating?
For P005.7: What attenuation (dB, 100 MHz-10 GHz) does a cheaply welded 1 mm steel pod enclosure with double-right-angle conduits achieve (target >= 62 dB)?
For P005.8: At what separation distance does EM/acoustic side-channel leakage between adjacent racks fall below a stated threshold?
* 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