Artificial intelligence (AI) is shaping charitable donations more and more. That means traditional theories about human fundraisers might not apply when AI is involved. In charitable storytelling, fundraisers use first-person or third-person narrative perspectives to convey their messages. Different narratives have varying effects on audiences' donation intentions, yet the most suitable narrative perspective for AI fundraisers remains unexplored. Drawing upon narrative transportation theory (NTT), we ran three studies exploring the interaction between narrative perspective and fundraiser identity (AI vs. human) in shaping people's intentions to donate. Study 1 found that people were more willing to donate money for first-person narratives from AI but third-person narratives from humans. Study 2 found evidence that this difference happens because of empathic concern. The different narrative perspectives influence people's empathic responses, which influence their donation intentions. Study 3 found that the identifiability of the person seeking help is a boundary condition. When the help-seeker's identity is ambiguous, the advantages behind first-person and third-person narratives disappear. This research extends the application of NTT to the field of human-computer interaction. The results provide empirical evidence that nonprofits can use to optimize their use of AI in fundraising.