A Standards Body for AI Integrity
AIO is a global non-profit that standardizes and operates a common vocabulary so the value, evidence, and source hierarchy followed by AI responses can be recorded, verified, and overseen.
We define AI responses by a value, evidence, and source hierarchy
AIO — the AI Integrity Organization — defines and operates the AIO 20002, which records, in a single line of standardized log, which value, which evidence, and which source an AI system prioritized when it reached a decision.
With this record, ethicists, clinicians, patients, and regulators can examine the basis of a decision through the same vocabulary, and the post-hoc review of a single case and the macro-level monitoring of millions of responses become possible in the same units of measurement.
We have values, but no vocabulary to express them
AI governance today is fragmented. Ethics committees write principle documents, engineering teams write guardrail code, and users only receive the outcome. At each stage, values are translated into a different language, and the loss in translation is never measured anywhere.
AIO does not answer which value is right. Instead, it defines a standard vocabulary for recording the relationships between values — which one takes precedence over another. With this vocabulary, a debate in the meeting room becomes a system configuration, a system response becomes proof a user can verify, and the accumulation of responses becomes a monitoring signal for regulators.
The four principles AIO follows
Trust must be measurable
A binary question of whether AI is "safe" or "aligned" does not build social trust. Trust only works when three hierarchies become visible as units of measurement: which values were prioritized, which evidence was prioritized, and which sources were prioritized.
The standard must be agreed before the values
Even when we cannot agree on the values themselves, we can agree on a vocabulary for expressing and comparing them. AIO Framework does not define which value is right; it defines how the hierarchy between values is recorded and operated.
Accountability that cannot be verified does not exist
Regulators, compliance teams, users — anyone must be able to verify the basis of a decision using the same vocabulary. Accountability must not be a slogan, but infrastructure that can be traced through a single line of log.
AI must serve diverse people
The assumption that any single value hierarchy is right for everyone is false. AIO Framework provides a vocabulary that lets users realign AI responses to their own value hierarchy, or move to an AI that is aligned with them.
Non-profit, neutral, international
AIO is a non-profit association (Swiss Verein) registered in Geneva, Switzerland, and is not beholden to any particular nation, company, or ideology. Its operational governance, quarterly consensus process, and financial operations are all designed to be traceable through public documents. Details are being updated progressively on the operational and financial transparency page.