AI-generated stories repeatedly invent non-existent lighthouse keeper Elias Thorne
A study by Cornell University researchers has uncovered why artificial intelligence repeatedly invents a non-existent lighthouse keeper named Elias Thorne, revealing a systemic pattern in AI-generated storytelling that is reshaping public perception. The phenomenon, documented in a paper published on arXiv on 26 June 2026, shows that large language models are trapped in a feedback loop where they reinforce the same names, professions, and settings—including faro (lighthouse) and farero (lighthouse keeper)—across more than 88% of generated stories.
Analyzing over 20,000 AI-written narratives, the team found that Elias appears in more than a quarter of all stories, despite being virtually absent from human literature. Other recurring elements include professions like watchmaker and librarian, and settings such as lighthouses, which appear in over half of the texts. The researchers ruled out human influence, concluding that the repetition stems from the training process itself. “The alignment process restricts content to safe, generic options,” said one co-author, “and user feedback loops reinforce these choices, creating a mode collapse where diversity collapses into formula.”
The consequences extend beyond fiction. According to 404 Media, the name Elias Thorne has now migrated from AI-generated stories into real-world platforms, where it is cited as the author of books, musician, and content creator. Software engineer Daniel May reported a surge in online searches for the name in 2026, indicating that AI’s synthetic output is being accepted as factual. This cycle is exacerbated by datasets that increasingly include machine-generated content, which, when re-ingested during training, deepens the repetition.
The result is a narrowing of narrative possibility. “Stories may appear varied on the surface,” the Cornell team noted, “but they all orbit the same safe, sanitized points.” Without intervention in data sourcing or training methods, the uniformity risks becoming self-perpetuating, turning AI storytelling into a hall of mirrors where every tale echoes the last.
Follow us for live European news
- 3
- 3
- 2
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
2 further sources not geolocated





:format(jpg)/f.elconfidencial.com%2Foriginal%2Fedc%2F4c2%2Faf6%2Fedc4c2af6798ceeb98d2edfa1d0791cc.jpg)

