P063 Composing trust across parties and components

Can a mutually secure verification system be built from unilaterally trusted components, and can the combination of trust roots be proved sound?

Composing trust from unilaterally trusted components.

Status

Analyzed Design and a stated soundness condition only, from a self-deprecated working draft (June 2026; treat as dated): a mutually secure auditing environment from three unilaterally trusted, independently sourced trusted computing bases; the February 2026 workshop report says the approach is 'designed to work with great power adversaries without trusting either side's chips', with specifics 'TBA'. Nothing built. The 'proved sound' clause has no work: AI 2040 calls proving the combination sound 'the open problem'. Hausenloy & Li's 'never trustless, only low-trust' is a position.

The security of redundant cross-checking hinges on the different devices not having the same vulnerabilities, which means they need to be built from independently sourced components, and not share (too much) code or architectural similarity in general.
Cankaya (2026), A System Overview for Near-Term, Low-Trust AI Compute Verification, version 0.2 working draft (LessWrong, archived revision 1.4.0) (2026-06-24), L489

Posed by

Also raised in: Petrie & Aarne (2025), Technical Options for Flexible Hardware-Enabled Guarantees (flexHEG Part 2); Cankaya (2026), A System Overview for Near-Term, Low-Trust AI Compute Verification (v0.2); Cankaya (2026), All Hands on Deck to Build the Datacenter Lie Detector; Hausenloy & Li (2026), The Compute Verification Post; Rahman (2026), Does Distributed Training Undermine Compute Governance? (MIRI); Wasil, Reed, Miller & Barnett (2024), Verification Methods for International AI Agreements.

Sub-questions

IDSub-questionWhat the sources report
P063.1For a stated evasion scenario, what is the combined detection probability of a proposed method set ('Swiss cheese' / multiple independent data sources), and which methods are correlated?open
P063.2Can international verification be trustless, or only low-trust?a source suggests an answer (not checked)

Sub-questions not yet stated precisely enough to check

IDSub-question
P063.3Can a mutually secure verification system be built from unilaterally trusted components, without trusting either side's chips, and can the combination of trust roots be proved sound?
P063.4Do redundant cross-checking TCBs from independent supply chains actually lack shared vulnerabilities?

Possibly relevant work found by automated search

Searched; nothing found for these sub-questions.

Data