R0043/2026-04-01/Q002 — ACH Matrix¶
Matrix¶
| H1: Direct requirements exist | H2: No requirements exist | H3: Indirect coverage only | |
|---|---|---|---|
| SRC01-E01: EU AI Act names automation bias | + | -- | ++ |
| SRC02-E01: NIST addresses confabulation not sycophancy | - | + | ++ |
| SRC03-E01: SR 11-7 effective challenge | + | -- | ++ |
| SRC04-E01: FDA human factors without sycophancy | - | + | ++ |
| SRC05-E01: IEEE 3119 without sycophancy | - | + | + |
| SRC06-E01: No explicit sycophancy regulation | -- | ++ | ++ |
Legend:
- ++ Strongly supports
- + Supports
- -- Strongly contradicts
- - Contradicts
- N/A Not applicable to this hypothesis
Diagnosticity Analysis¶
Most Diagnostic Evidence¶
| Evidence | Why Diagnostic |
|---|---|
| SRC06-E01 | Georgetown's explicit finding of "no explicit regulatory framework" eliminates H1 and confirms the gap |
| SRC01-E01 | EU AI Act naming "automation bias" eliminates H2 (total gap) while supporting H3 (indirect coverage) |
Least Diagnostic Evidence¶
| Evidence | Why Non-Diagnostic |
|---|---|
| SRC05-E01 | IEEE 3119 is a procurement framework; its absence of sycophancy provisions is unsurprising |
Outcome¶
Hypothesis supported: H3 — Indirect coverage exists through domain-specific mechanisms, but the specific model behavior is not named or directly targeted.
Hypotheses eliminated: H1 — No direct sycophancy requirements found. H2 — Indirect coverage exists, so a total gap is overstated.
Hypotheses inconclusive: None.