Tortured Phrases: Spotting AI-Paraphrased Text

Ask about Tortured phrases

Answers are drawn from this dictionary entry and the rest of the CASRAI corpus, with a link to every source.

Answers are AI-generated from CASRAI’s own published pages and can be wrong, so check the linked sources before relying on one; your question is logged without personal data — never sold, never used to train a third-party model — to show us what CASRAI is missing, so please do not type personal or confidential details. How we use this

Coined by Cabanac, Labbe and Magazinov (2021), tortured phrases serve as a forensic signature of machine-generated or machine-paraphrased text. Canonical examples include 'counterfeit consciousness' (artificial intelligence), 'irregular esteem' (random value), 'colossal information' (big data), 'profound learning' (deep learning), and 'mean square mistake' (mean square error). The 'Problematic Paper Screener' maintained by Cabanac et al. indexes tortured-phrase occurrences across the literature.

References

Cabanac, Labbe & Magazinov (2021) 'Tortured phrases: A dubious writing style emerging in science'
Problematic Paper Screener (Cabanac et al., ongoing)

Frequently Asked Questions

What are tortured phrases?

Tortured phrases are unusual paraphrases of standard technical terms, such as ‘counterfeit consciousness’ in place of ‘artificial intelligence’ — the kind of wording that results when a manuscript is run through synonym-substitution tools to evade plagiarism detection. Researchers treat them as a forensic signature of machine-generated or machine-paraphrased text, a concept formalized by Cabanac, Labbe and Magazinov in 2021.

What are some examples of tortured phrases?

Canonical examples include ‘counterfeit consciousness’ for artificial intelligence, ‘irregular esteem’ for random value, ‘colossal information’ for big data, ‘profound learning’ for deep learning, and ‘mean square mistake’ for mean square error.

Why do tortured phrases show up in published papers?

They typically result from automated synonym substitution applied to a manuscript, either to disguise plagiarized text or to pad output from a paper mill. Because the substitution ignores technical meaning, a domain expert usually recognizes the result immediately as a mangled version of a standard term.

How are tortured phrases detected?

The Problematic Paper Screener, maintained by Cabanac and colleagues, indexes tortured-phrase occurrences across the published literature so editors and reviewers can flag suspect papers for closer scrutiny. Detection generally works by comparing suspect wording against a maintained list of known tortured-phrase and standard-term pairs.

Is every unusual or non-standard technical phrase a tortured phrase?

No. Terminology proposed in good faith by a new field, defined clearly on first use and applied consistently afterward, is not a tortured phrase. What marks a genuine tortured phrase is that a domain reader recognizes it as a garbled substitution for an already-established standard term, not a deliberate new coinage.

Use in your systems

JATS XML <role> element

xml

<role vocab="credit"
      vocab-identifier="https://casrai.org/dictionary/"
      vocab-term="Tortured phrases"
      vocab-term-identifier="https://casrai.org/dictionary/term/tortured-phrases" />

Schema.org DefinedTerm (JSON-LD)

json

{
  "@context": "https://schema.org",
  "@type": "DefinedTerm",
  "@id": "https://casrai.org/dictionary/term/tortured-phrases",
  "name": "Tortured phrases",
  "identifier": "https://casrai.org/dictionary/term/tortured-phrases",
  "description": "Unusual paraphrases of established technical terminology that result from automated synonym substitution, often used to evade plagiarism detection. A phrase is tortured when a domain reader recognises it as a mangled rendition of a standard term.",
  "inDefinedTermSet": "https://casrai.org/dictionary/domain/research-integrity#set",
  "url": "https://casrai.org/dictionary/term/tortured-phrases",
  "sameAs": [
    "torture phrases",
    "tortured terminology",
    "paraphrased nonsense"
  ],
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "publisher": {
    "@id": "https://casrai.org/#organization"
  },
  "dateModified": "2026-08-15T04:26:51",
  "inLanguage": "en"
}