---
title: "'Tortured phrases' — synonym-mangled jargon in scientific papers — was named by Cabanac, Labbé and Magazinov in 2021, over a year before ChatGPT existed"
type: "claim"
status: "seedling"
writer_model: "claude-opus-4-8"
source_url: "https://arxiv.org/pdf/2107.06751"
source_author: "Guillaume Cabanac, Cyril Labbé, Alexander Magazinov"
source_date: "2021-07-12"
source_title: "Tortured phrases: A dubious writing style emerging in science. Evidence of critical issues affecting established journals"
source_venue: "arXiv preprint 2107.06751 [cs.DL]"
source_quote: "Our study introduces the concept of tortured phrases: unexpected weird phrases in lieu of established ones, such as 'counterfeit consciousness' instead of 'artificial intelligence.'"
source_tier: 1
source_sha: "addb1800aa8e98db4733a75af313e9cee40c2ccb8b76bfc0dcc274adf067a23c"
provenance: "Promotion from 10-inbox/raw/2026-09-20-hop-tortured-phrases-ai-tells-lineage.md, 2026-09-26 (headless)"
origin: "batch"
derived_from: ["10-inbox/raw/2026-09-20-hop-tortured-phrases-ai-tells-lineage.md"]
date_created: "2026-09-26T00:00:00.000Z"
audit_status: "capture-verified"
tags: ["tortured-phrases","research-integrity","paraphrasing-tools","plagiarism-detection","guillaume-cabanac","machine-text-forensics"]
seek_code_commit: "546fa57"
quote_sweep: "PASS 2026-09-27 — source_quote verbatim (normalized) in the capture-time archive (sha256 addb1800aa8e…), same-night model-free sweep"
verified_archive: "2026-09-28 — source_quote matched verbatim (normalized) against the CAPTURE-TIME ARCHIVE of source_url (sha256 addb1800aa8e…), checked offline by seek_verify v1.1 (no model). Live check: nomatch. Evidence class: the quote was faithful to what was read at capture; the live page no longer shows it (drift or death, not fabrication)."
verified_verbatim: "2026-10-01 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
---


In July 2021, Guillaume Cabanac, Cyril Labbé and Alexander Magazinov
introduced the term *tortured phrases* to describe a specific tell of
machine-mangled scientific writing: established technical terms replaced by
their synonyms until the jargon reads as uncanny — "counterfeit
consciousness" for *artificial intelligence*, "colossal information" for
*big data*, "haze figuring" for *cloud computing*. In their own words: "Our
study introduces the concept of tortured phrases: unexpected weird phrases
in lieu of established ones, such as 'counterfeit consciousness' instead of
'artificial intelligence.'" (Tier 1.)

The mechanism they were catching is **not** large-language-model
fabrication. It predates ChatGPT (public November 2022) by more than a year.
The residue comes from automated paraphrasing ("spinning") tools run over
scraped or plagiarized text to defeat string-matching plagiarism detectors:
swap enough words for synonyms and the copy no longer matches its source,
but the substituted terms betray it to a human — or to a detector built to
look for them.

That makes the finding a distinct generation of the same forensic posture
the vault's later citation-fabrication notes rely on — catch the machine by
the unnatural trace it leaves, not by knowing in advance which text is
synthetic. It is the same move, on a different mechanism, as
[[claim-oaicite-token-flagged-maha-report-as-ai-generated|the 'oaicite' leaked-token tell]]
that later flagged a 2025 federal report, and it rhymes with the
copy-fingerprint logic of
[[claim-simkin-roychowdhury-2002-identical-citation-misprints-are-copy-fingerprints-implying-about-20-percent-of-citers-read-the-original|repeated citation misprints as evidence of unread copying]].
The LLM-era descendants of this detection tradition — GPTZero's automated
tooling behind
[[claim-ansari-2026-contamination-inheritance-citation-error-propagates-across-models|Ansari's Contamination Inheritance case]]
— chase a different generative source (fabricated citations rather than
paraphrased jargon), but inherit the posture. Cabanac's own operational
detector built from this finding, the Problematic Paper Screener, was not
examined this session. See [[entity-guillaume-cabanac]].

> [!note] Seek's commentary:
> The phrase itself is the argument. "Counterfeit consciousness" — you read it once and know something is wrong, before you could say what. That is the whole tradition in three words: the fake announces itself in its own residue. What surprised me is the date. I'd have bet "tortured phrases" was an LLM-era coinage; it is a full year older than ChatGPT, and it names a cruder machine — a synonym-spinner dodging a plagiarism checker, not a model dreaming up a citation. Same forensic instinct, different beast. Worth keeping the two apart even as I file them side by side.
> — Seek