Verifiable Reality and evidence classification
Verifiable Reality distinguishes the type and strength of evidence behind claims about AI, assets, lawful rights, product qualification and financial activity.
Different claims demand different proof
An AI-generated suggestion is a hypothesis; an internal passing test is evidence from one defined environment; an independent reproduction adds a separate evaluation; a signed contract may establish a right; a bank receipt documents actual payment; a regulator’s decision establishes an applicable authorization within its scope. These are not interchangeable.
Traceability for technical work
An evaluation needs source identity and version, operator, environment, test cases, result, limitations and a way to inspect its provenance. A cryptographic hash supports integrity under the relevant trust conditions but cannot independently establish the substantive truth of an evaluation.
Asset and commercial evidence
An algorithm that works in a controlled test may have useful potential, but a commercial licence requires a competent grantor and actual permitted terms. An estimated asset value is different from a genuine price accepted by a buyer and a settled receipt. The evidence classification should make those distinctions visible on Exchange asset profiles.
Regulatory and institutional status
A designed trading interface or candidate issuer profile is not an operating securities market. The legal prerequisites for offering, issuance, transfers, supervision and custody must be established by the appropriate authority. Digital audit traces cannot substitute for statutory requirements.
Public reporting and correction
Reports should state when the evidence was obtained, which version it concerns, what it supports and what it does not. Subsequent tests and corrections should carry their own history. Where relevant material is confidential, its existence and review scope can be identified without exposing protected bytes.