R0029/2026-03-27/Q001/H1¶
Statement¶
Multiple formal frameworks, standards, or structured proposals for AI attribution have been published by credible institutions, representing an established (if fragmented) field.
Status¶
Current: Partially supported
The evidence shows that structured proposals exist — the IBM AI Attribution Toolkit, the AIA icon system (Avery, Abril & del Riego), and the CHI 2025 research on attribution perceptions all represent formal work. However, calling them "established" overstates the situation. None has achieved widespread adoption, and no standards body has formally adopted an AI-specific attribution standard.
Supporting Evidence¶
| Evidence | Summary |
|---|---|
| SRC01-E01 | CHI 2025 paper demonstrates structured research on attribution perceptions with 155 participants |
| SRC02-E01 | IBM released a working AI Attribution Toolkit in May 2025 with structured questionnaire approach |
| SRC03-E01 | AIA icon system proposed in Northwestern JTIP for legal and ethical AI disclosure |
Contradicting Evidence¶
| Evidence | Summary |
|---|---|
| SRC04-E01 | CRediT taxonomy, the closest existing standard, has not been formally extended to cover AI contributions |
Reasoning¶
While multiple proposals exist, they have not coalesced into a recognized standard. The IBM toolkit is explicitly described as "a first pass" at what a voluntary reporting standard might look like. The AIA icon system is a law review proposal, not an adopted standard. The CRediT taxonomy — the only NISO-backed attribution standard — does not yet accommodate AI. This supports the hypothesis partially: proposals exist, but "established" is too strong.
Relationship to Other Hypotheses¶
H1 is partially correct but overstates the maturity. H3 more accurately characterizes the current state. H2 is eliminated because structured proposals clearly go beyond binary disclosure.