Humanized but Still True? Auditing Facts and Detection in AI-Text Humanizers

All 11 commercial AI-text humanizers on a public leaderboard changed or dropped at least one checkable unit — numbers, dates, person and organization names, or quoted spans — in 35-65% of the October 2026 documents that contained one, according to an arXiv paper (2611.00359v1) that audited fact retention with a rule-based checker validated by two AI annotators. Counting numbers and names alone, 10 of the 11 tools still altered at least one in 20% or more of documents, and the September 2026 cycle was similar at 19-50%. The authors also report that a general-purpose decision model, Jev, separated human originals from humanized rewrites with AUROC 0.864 (0.827-0.902), 0.18-0.22 above Binoculars, Fast-DetectGPT and RADAR, while flagging 24% of human originals as AI at a 0.5 threshold; the checker and per-item results are released.

arXiv:2611.00359v1 Announce Type: new Abstract: AI "humanizers" rewrite machine-generated text so that detectors classify it as human-written. Public leaderboards score evasion and embedding similarity, but not whether checkable details survive. We check whether numbers, dates, person and organization names, and quoted spans survive each rewrite, using a rule-based checker validated by two AI annotators precision 0.84-1.00, recall 0.93-0.95; no human annotation . In the October 2026 cycle of a public humanizer leaderboard, all 11 commercial tools changed or dropped at least one of these units in 35-65% of the documents that contained one. Most changes were to quoted spans; counting numbers and names alone, 10 of the 11 tools still altered at least one in 20% or more of the documents. The September 2026 cycle was similar 19-50% . An open 12B humanizer kept at least 97% of numbers and names in fact-dense U.S. federal agency text, but reworded 67-78% of the quoted spans, against its own prompt. Unit retention and detector evasion were not clearly related across tools Spearman rho = -0.46, 95% CI -0.65 to 0.08 , and the open model's four released files two of them quantization-aware low-bit files did not show the pre-registered monotone trade-off. On federal texts with matched human, AI-styled and humanized versions, a general-purpose decision model, Jev, asked a single zero-shot question, separated human originals from humanized rewrites with AUROC 0.864 0.827-0.902 , 0.18-0.22 higher than Binoculars, Fast-DetectGPT and the adversarially trained RADAR, while calling 24% of the human originals AI at a 0.5 threshold. On the AI-styled rewrites by 2026 models, both perplexity-based detectors scored below chance AUROC 0.37 and 0.39 , also with line breaks removed. Hypotheses, data rules and analyses were frozen before the main runs; two earlier data contacts are disclosed. We release the checker and the per-item results.