E01¶


Research	R0029 — Plural Voice Attribution
Run	2026-03-27
Query	Q003
Source	SRC05
Evidence	SRC05-E01
Type	Factual

NeurIPS LLM policy on authorship and disclosure

URL: https://neurips.cc/Conferences/2025/LLM

Extract¶

FACT: NeurIPS states "Only humans are eligible to be authors." Authors are "fully responsible for all the content in their paper, including text, figures, and methodology, regardless of what tools (e.g., LLMs) they have used." Disclosure required only when AI tools are "part of the methodology." Grammar/editing use does not require disclosure. NeurIPS reserves the right to investigate and revoke publication status for violations.

Relevance to Hypotheses¶

Hypothesis	Relationship	Strength
H1	Supports	Prohibition confirmed
H2	Contradicts	Formal policy exists
H3	Supports	Methodology-only disclosure threshold differs from ACM/IEEE acknowledgment requirement

Context¶

NeurIPS takes the most permissive approach among ML conferences — disclosure only if AI is part of research methodology, not for writing assistance. This contrasts with ACM's requirement to disclose all generative AI use.