Q001 — ACH Matrix¶


Research	R0044 — Expanded Vocabulary Research
Run	2026-04-01
Query	Q001

Matrix¶

	H1: Enforceable system-side requirements exist	H2: Partial/emerging requirements exist	H3: No requirements at all
SRC01-E01: CSET three-tiered framework identifies technical/design level	+	++	--
SRC02-E01: EU AI Act Article 14 design requirements for automation bias awareness	+	++	--
SRC03-E01: NIST AI 600-1 identifies confabulation and overreliance risks	+	++	--
SRC04-E01: FDA CDS guidance focuses on human independent review only	--	+	++
SRC05-E01: FINRA focuses on supervisory procedures only	--	N/A	++
SRC06-E01: FAA/EASA focus on certification/testing only	--	+	++

Legend: - ++ Strongly supports - + Supports - -- Strongly contradicts - - Contradicts - N/A Not applicable to this hypothesis

Diagnosticity Analysis¶

Most Diagnostic Evidence¶

Evidence	Why Diagnostic
SRC02-E01	EU AI Act Article 14 is the single most discriminating piece of evidence — it eliminates H3 (some system-side provisions exist) while simultaneously preventing H1 from being supported (provisions focus on transparency, not output behavioral constraints)
SRC04-E01	FDA's relaxation of multiple alternatives requirement is highly diagnostic — it shows the most relevant sector moving away from a system-side mechanism

Least Diagnostic Evidence¶

Evidence	Why Non-Diagnostic
SRC05-E01	FINRA's supervisory focus is consistent with both H2 and H3, providing little discrimination between them

Outcome¶

Hypothesis supported: H2 — Requirements exist in partial/emerging form. System-side provisions exist (EU AI Act Article 14, NIST AI 600-1) but focus on transparency and risk identification rather than constraining AI output behavior.

Hypotheses eliminated: H1 — No enforceable system-side requirements constraining AI output behavior were found. H3 — EU AI Act and NIST provisions demonstrate the regulatory landscape is not entirely silent on the system side.

Hypotheses inconclusive: None.