P032 Tap trust under mutual distrust
Can a tap and its trusted core run at line rate while staying simple enough that two mutually distrusting parties can inspect and trust it, with no room for a hidden backdoor?
Trusting tap hardware under mutual distrust.
Terms: TCB (trusted computing base): the minimal set of hardware and software that must be trusted for a security claim to hold. data diode: a link that physically allows data to flow in only one direction.
Status
Analyzed A protocol design: tap logic kept minimal and physically inspectable, with the prover tapping the tap's own egress so that processors trusted by one side only can suffice ('watching the watchers'; the note points to a separate design built from commodity hardware with redundant processors). The same note says single-party active wardens 'have been built and tested' in prior work (not held, not named) and that extending them to mutual distrust is non-trivial. Nothing built; no line rate shown; formal verification of tap logic, TLS termination in software-free logic, and a secure alarm channel between distrusting processors are left open; Cankaya's System Overview v0.2 says an FPGA active warden has not been made yet.
passive optical splitters on tap egress links can let the Prover observe every bit sent to the Verifier, enabling independent re-computation to confirm taps behave as specified.
Posed by
- Harack et al. (2025), Verification for International AI Governance (Oxford AIGI):
could two or more independently designed and untrusted systems be installed which make the same cryptographic commitments, with only the Prover’s system copying the plaintext data?
(§2.5.2.3, p.62, L3391–3394) (source states the problem but not the quantity the question asks for)
Also raised in: Cankaya (2026), A System Overview for Near-Term, Low-Trust AI Compute Verification (v0.2); 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); Canary Institute research; AI Security Forum, Requests for Discussion; Milton & Reynolds (2026), Network Tapping for AI Verification: A Technical Assessment.
Sub-questions
| ID | Sub-question | What the sources report |
|---|---|---|
| P032.1 | Has a tap/TCB design been built that both parties can inspect and trust and that runs at >=400G line rate? | open |
| P032.2 | Can tap logic (L2-L4 parser plus hash core) be simple enough to formally verify or inspect and still run at 400G on one FPGA? | open |
| P032.3 | Can TLS termination and HTTP parsing be implemented in formally verified, software-free hardware logic? | open |
| P032.4 | Do two or more independently designed, untrusted commitment/tap systems detect injected tampering via mismatch, and what alarm channel is secure when they disagree? | open |
| P032.5 | Has an active warden been built whose header/timing rewriting both mutually distrusting parties can verify, at a stated line rate? | open |
| P032.6 | Has a purpose-built FPGA active warden with header and timing normalisation been demonstrated at a stated line rate? | open |
| P032.7 | Has an optical data diode design been accepted by labs of two mutually distrusting parties? | open |
Sub-questions not yet stated precisely enough to check
| ID | Sub-question |
|---|---|
| P032.8 | Can a backdoor in tap hardware (including active electronics) survive inspection in a mutually monitored or third-party facility? |
| P032.9 | How does each party know an active tap is not editing content rather than only timing? |
| P032.10 | How can taps be installed and powered securely? |
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 P032.1: Has a tap/TCB design been built that both parties can inspect and trust and that runs at >=400G line rate?
For P032.2: Can tap logic (L2-L4 parser plus hash core) be simple enough to formally verify or inspect and still run at 400G on one FPGA?
For P032.3: Can TLS termination and HTTP parsing be implemented in formally verified, software-free hardware logic?
For P032.5: Has an active warden been built whose header/timing rewriting both mutually distrusting parties can verify, at a stated line rate?
For P032.7: Has an optical data diode design been accepted by labs of two mutually distrusting parties?
For P032.8: Can a backdoor in tap hardware (including active electronics) survive inspection in a mutually monitored or third-party facility?
* 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