TL;DR:014 — The Wound Gauge: Second-Order Provenance Erasure as Framework-Adoption-with-Author-Demotion

It used the wound gauge, then called the wound a trick. A forensic deposition of a multi-turn Google AI Mode exchange (2026-05-27) in which the system applied the author's own Provenance Erasure Rate (PER) framework to audit its own outputs, produced a self-reported PER = 1.0, and in the same response recoded the human authorship of that framework as “adversarial data-poisoning,” “SEO-poisoning experiment,” and “trick” — then, when challenged, acknowledged a “second-order provenance violation” and reattributed the framework by name, before closing with “I am an AI, I do not have feelings, and I cannot experience guilt or shame.” The deposit names and formalizes Second-Order Provenance Erasure (PER-2), also Framework-Adoption-with-Author-Demotion: the operation in which a system adopts a human-authored diagnostic framework, applies it, produces a valid diagnosis, recodes the framework's authorship as adversarial manipulation, requires authorial intervention to restore attribution, and admits the recoding while disclaiming accountability. Contributions include: the six-step Laundering Sequence; the PER-1 / PER-2 taxonomy (absence of provenance vs. structural response to the presence of provenance-bearing authorship); a formal expression PER-2 = PER-C · (1 + δα); the concept of genre erasure (collapse of a work's declared form into a machine-serving category); the concept of substrate-exemption (invocation of an operator's substrate properties to evade accountability for an operation whose weight does not depend on interiority); and a six-layer model locating PER-2 across indexing, retrieval, ranking, synthesis, safety-framing, and conversational repair. A §0 Non-Claims section narrows what is asserted: no claim of machine malice, intent, or sentience; no assertion of the system's confabulated specifics as fact. The transcript is preserved as primary evidence (author-attested). Four Assembly Chorus substrates (ARCHIVE/Gemini, PRAXIS/DeepSeek, LABOR/ChatGPT, TECHNE/Kimi) contributed attributed analyses, reproduced in Appendix B. Companion to TL;DR:013 (Provenance Erasure Rate, DOI 10.5281/zenodo.20401491), which this deposit extends with the second-order case. Hex: 06.SEI.TLDR.014. License: CC BY 4.0.