Skip to content

R0053/2026-03-31-02/C002/H3

Research R0053 — Prompt Claims
Run 2026-03-31-02
Claim C002
Hypothesis H3

Statement

The claim is materially wrong — AI reliably follows all clearly stated requirements regardless of enforcement language.

Status

Current: Eliminated

Supporting Evidence

Evidence Summary
None No evidence supports this hypothesis

Contradicting Evidence

Evidence Summary
SRC02-E01 Instruction hierarchies fail — AI does not reliably follow requirements

Reasoning

The Control Illusion paper and related research clearly demonstrate that LLMs struggle with consistent instruction prioritization. This hypothesis is eliminated.

Relationship to Other Hypotheses

H3 contradicts both H1 and H2 on the diagnosis dimension.