This week, The Atlantic declared that the Granta AI fiction scandal marks a new phase in the struggle to keep literature human. Simultaneously, across town, the Met's Costume Institute opened a show that, per Artnet's review, gets literally under the skin, unpacking the interplay between fashion, the body, and art. The collision is not coincidental. Both events are circling the same anxious question: what requires an embodied human to produce, and what happens when that requirement dissolves?
Authenticity as the Last Luxury Good
The Granta scandal's real weight is not the individual case but the institutional response. As The Atlantic reports, the magazine's handling points to a new phase: not denial, not panic, but a kind of uncertain renegotiation of what authorship means. Fashion has been through this. The Met Costume Institute show's focus on skin and the body is a kind of counter-programming, a claim that the most irreducible unit of fashion is the body that wears it, not the design that covers it. Jordan Brand's new Triangle sneaker, designed for every position on the court, is marketed specifically around athletic embodiment. Even sneaker design, in 2026, is justifying itself through the body it serves.
The Cultural Verification Problem
What links literary AI fraud and fashion's body turn is a shared verification problem. A 2026 arXiv paper, Wang et al. on data probes for LLM performance, makes the technical version of this argument: we don't yet have adequate tools to understand what makes certain training data produce certain outputs. The same epistemic gap applies to prose. We cannot yet reliably detect AI writing because we don't fully understand what makes human writing human. Fashion's answer is to go back to the skin. Literature hasn't found its equivalent move yet. LaKeith Stanfield, currently trending for his chameleonic career and a role in I Love Boosters, gave an interview this week where he said: "say something that's worth something." It's an instruction that applies equally to novelists and large language models, and the fact that we can't tell which is which is the whole crisis.