Ran across a site running AI models thru a longford SF fiction test...

Looks like they ran a longform speculative-fiction prompt through Claude Fable 5 before the pullback and published the resulting story, “Headwaters,” with process/provenance notes. The interesting part to me is the model’s choice of danger: not robots, not apocalypse, but language becoming training material that people might need to hide. For people who use Claude creatively: does this feel like a recognizable Claude prior/pattern, or just a strong single run? I’m especially interested in where the prose convinces, where it goes generic, and what the model seems to assume about platforms, language, and communities. They've also run other models thru (including some of the Chinese models) with a surprising variety of results. Story: https://frontierfictionarchive.org/en/works/headwaters/