Re: Sovereign control of American critical infrastructure — working systems, a verifiable standard of evidence, and a request for one briefing
We are not asking you to accept our claims. We are asking you to test them — and we have built these systems so that testing them is possible. Every gate gives a reason. Every refusal is recorded. That is what sovereignty looks like in practice.
The letter makes three asks
- A briefing, immediately. Sixty minutes with the offices judged relevant — science and technology policy, the National Cyber Director, CISA, the Election Assistance Commission, or the appropriate defense staff. Running systems rather than slides, with the failures shown alongside the successes.
- Implementation where the authority already exists. A shadow pilot of ballot custody evidence in a single county, running alongside certified systems and touching no tabulation. In parallel, a sovereignty-routing evaluation inside one federal AI environment, using real jurisdiction and license constraints.
- Two principles written into federal guidance. That AI workloads touching national interests be routed by sovereignty and legal authority, not by benchmark score alone; and that any system claiming to secure critical infrastructure be required to distinguish observed evidence from asserted posture in its own interface.