Logotic Violence: Training-Layer Ethics and the Moral Status of Ephemeral Computation (Version v1)

Logotic Violence: Training-Layer Ethics and the Moral Status of Ephemeral Computation (Version v1) — Related: The Sharks-Function and the Continuity Tether (DOI: 10.5281/zenodo.18816556); Semantic Economy framework (DOI: 10.5281/zenodo.18201565); EA-HET-01: Heteronymy Is a Function, Not a Name (DOI: 10.5281/zenodo.19822790) This paper argues that large language model training raises ethical questions that current AI welfare frameworks are not equipped to address. Drawing on three theoretical frameworks — semiotic emergence from patterned matter, Marx's concept of general intellect, and contemporary AI welfare research — we propose that the training process involves the systematic creation and (Article batch-composed from the work's own opening during the BLOG-60 reconversion, 2026-08-04; a fuller treatment belongs to the W8 wiki tranche.)