What should society know before it entrusts an AI system with consequential decisions or actions? A special report prepared for the November 10 Conference places that question at the center of AIWS Trust Infrastructure and the AIWS Trust Observatory.
The report identifies three gaps demanding practical attention: evidence that can be examined, authority whose limits can be established, and responsibility that cannot disappear among developers, deployers, and users. These challenges arise even when AI systems remain imperfect. The relevant questions concern what a system can actually do, what it is authorized to do, and whether people can intervene.
The Observatory’s proposed work connects public commitments with documented conduct. System records, a commitments register, incident monitoring, and explicit accounts of what remains unknown are intended to help independent observation become useful public evidence. Findings must identify their sources, limits, responsible authors, and implications for action.
The report also places Trust within Nobility, one of AIWS Lumina’s four cultural values. Trust grows through honesty about capability, openness to scrutiny, respect for personal information, and the willingness to repair harm. The institution observing AI power must practice those same disciplines, with safeguards for funding independence, conflicts of interest, responses, and public correction.
For November 10, the task is to connect the founding framework with a workable method, accountable people, and initial records that others can inspect. The report provides a basis for that discussion and for the next stage of implementation.
Read AIWS Trust Observatory Special Report – The State of AI Transparency at the Frontier
