Study of 13,867 Disclosure Statements Exposes Gaps in AI Disclosure Practices

A study accepted to EMNLP 2026 Findings, "Expectations and Practices around AI Disclosure in CS Research," systematically examined AI-use disclosure in computer science papers: although policies are now widespread, they remain highly vague and outdated, and actual disclosure practices diverge sharply from academic expectations.

Confirmed

• The study analyzed 13,867 disclosure statements from EMNLP 2025 and ICLR 2026.
• CS researchers consider disclosure most essential for research-design tasks with low human involvement (e.g., hypothesis generation, synthetic data), and least necessary for writing and reporting tasks.
• Practice mismatches expectations: writing assistance is deemed less necessary to disclose yet is the most frequently mentioned.
• AI-use policies, while ubiquitous, are highly vague and largely unchanged since 2023, failing to keep pace with rapidly evolving AI tools.
• The analysis found 95 distinct ICLR 2026 papers using identical disclosure text; after ruling out shared authors or institutions, the source remains unknown.
• The authors propose a task taxonomy separating mandatory from optional disclosure, plus standardization recommendations.

Unconfirmed

• The origin of the 95 identical statements remains a mystery, as poster @danish037 notes.

Why it matters

• Vague and stale disclosure norms undermine academic integrity and the auditability of AI use; the study offers large-scale empirical evidence for venues revising policies.
• The 95 duplicate statements suggest the disclosure process may be treated perfunctorily via template copying, underscoring the need for mandatory, standardized disclosure.

2026-08-25 ~ 2026-08-25 · 6 related posts

• Research Finds AI Use Policies Remain Vague and Under-Specified — danish037 · 2026-08-25
• Study Reveals Disconnect Between Expectations and Practice in AI Disclosure — danish037 · 2026-08-25
• [source] Study analyzes AI disclosure statements in 13k+ papers, finds disconnect — danish037 · 2026-08-25
• [source] Mystery: 95 unique ICLR 2026 papers share the exact same disclosure statement — danish037 · 2026-08-25
• [source] Study of 13,867 AI Disclosure Statements Finds Policies Vague and Practices Misaligned — danish037 · 2026-08-25

1 near-duplicate retellings: danish037