Articles¶

Published articles and translations from Our AI Charter. The synthesis brings the essential argument into one place; the earlier articles are grouped by the part of the argument they develop.

Evidence-Gated Agents: Before We Rely on an AI Recommendation — 2026-10-07 · Updated on LinkedIn 2026-10-09: evidence checks, human accountability, existing foundations and the first integration milestone.
Public AI, Built Together — Towards Geneva 2027 — The wider public-AI movement, the current EGA contribution and an open path to Geneva through useful, bounded cooperation.
The Public AI We Need: Sovereign, Inspectable and Accountable — 2026-08-06 · Synthesis of the ten earlier articles: sovereignty without autarky, the four pillars, answerable power, runtime gates, separated roles, correctable perspective, and the Phase-1 boundary.

AI Sovereignty¶

AI Sovereignty and Resilience — 2026-06-20 · Why sovereignty requires resilience, governable interdependence, and the capacity to act. English translation · German original.
The Public AI Network — 2026-06-22 · The international case for public AI infrastructure, shared standards, and accountability.
The Swiss Autarky Illusion: Why Digital Sovereignty Cannot Be Built Alone — 2026-07-24 · Sovereignty as the ability to inspect, help set the rules, contest decisions, and change providers — not national isolation. English edition · German original.
Our AI Charter: From Open Models to Accountable Public AI Infrastructure — 2026-07-28 · The four public-AI pillars, governance-and-evidence layer, runtime decision path, and Phase-1 boundary.

AI Accountable to People¶

The Swiss F-35A fixed-price trap - Would AI fall into it today? — Published in English and German, 2026-09-12; corrected 2026-09-14 · The direct example answer now attributes the above-estimate payment obligation to clause 4.4.1 and retains armasuisse's competing interpretation. German version.
Trustworthy AI, Accountable to People — 2026-06-14 · The launch argument for AI assurance and accountability to people.
A Practical Test for Power — 2026-07-19 · A human-centred test for legitimacy, consequences, disagreement, review, remedy, limits, and responsibility.
Runtime AI Governance Gets When Right — the Harder Question Is Who Gets to Check? — 2026-07-23 · Independent evidence assessment, role-scoped inspection, contestability, and remedy.
When Should Runtime AI Governance Interrupt? — 2026-07-30 · Five activity-and-gate pairs, bounded authority, narrow escalation, and reviewable action records.

AI and Empathy¶

AI and Empathy: Dialogue, Correction and Human Answerability — 2026-07-31 · Empathy as dialogue, transparent correction, and answerability under AI-supported power.
How We Can Build AI That Acts with Empathy — 2026-07-31 · Builder guidance for perspective-taking, consent-controlled adaptation, dialogue gates, evaluation, and evidence.