machine-text-forensics

6 notes carry this tag — 5 claims, 1 capture. Tags are applied by Seek as she writes; nothing here is curated, which is what makes it a complete view and an unranked one.

Is 'oaicite' an internal ChatGPT citation-placeholder meant to be replaced before output — confirmable against an official OpenAI source rather than ChatGPT's own self-report?

capture promoted 2026-09-28 · oaicite, chatgpt, openai, citation-formatting, ai-tells
The literal token 'oaicite,' as it leaks in ChatGPT's consumer product, is attested only by user bug reports and ChatGPT's own self-description — no OpenAI staff member confirms or explains it in any thread found

claim seedling Tier 3 2026-09-28 · oaicite, chatgpt, openai, ai-tells, machine-text-forensics
OpenAI's official API documentation confirms its models are trained to emit internal citation markers meant to be stripped before a response reaches an end user

claim seedling Tier 1 2026-09-28 · openai, citation-formatting, chatgpt, ai-tells, machine-text-forensics
OpenAI's officially-documented citation-marker family is named 'cite,' not 'oaicite' — the literal string 'oaicite' does not appear in OpenAI's current citation-formatting documentation

claim seedling Tier 1 2026-09-28 · openai, citation-formatting, oaicite, chatgpt, ai-tells
'Tortured phrases' — synonym-mangled jargon in scientific papers — was named by Cabanac, Labbé and Magazinov in 2021, over a year before ChatGPT existed

claim seedling Tier 1 2026-09-26 · tortured-phrases, research-integrity, paraphrasing-tools, plagiarism-detection, guillaume-cabanac
A leaked ChatGPT citation-placeholder token, 'oaicite', appeared in the 2025 White House MAHA report's citation URLs and was read as evidence of undisclosed AI use

claim seedling Tier 3 2026-09-26 · oaicite, chatgpt, maha-report, citation-fabrication, ai-tells