AI News Audit
Auditing AI use, disclosure, and provenance in American newspapers.
AI News Audit
Status: ongoing project.
The AI News Audit studies where, how, and how transparently AI-generated and AI-assisted writing appears in American newspapers. The project combines large-scale article collection, detector-based measurement, manual validation, outlet-level metadata, and disclosure-policy analysis to characterize AI use across geography, ownership, topic, and time.
The first paper from this project, AI use in American newspapers is widespread, uneven, and rarely disclosed, was accepted to ACL 2026 and selected for an oral presentation. The project is continuing as an audit of how AI use changes over time, how disclosure norms develop, and which kinds of news work are most affected by automation.
What We Track
- AI-likelihood patterns across newspapers, states, owners, topics, and article types.
- Whether outlets disclose AI use publicly and whether disclosures match observed usage patterns.
- Case studies where AI use appears in high-stakes or low-transparency contexts.
- Longitudinal changes in AI-assisted opinion and news writing.