AI for Researchers · Visual Deck
What publishers actually require, why detectors cannot enforce it, and how to disclose correctly
{{Presenter Name}}
Landscape as of August 2026
2-Minute Recap
flowchart LR s1(["1 · Foundations
the map · the taxonomy · CRIT"]):::done --> s2(["2 · Literature
tool matrix"]):::done s2 --> s3(["3 · Reading
summary-trust triage"]):::done s3 --> s4(["4 · Writing & integrity
today"]):::today s4 --> s5(["5 · Data
& code"]):::rest s5 --> s6(["6 · Multi-
disciplinary"]):::rest s6 --> s7(["7 · Discipline
deep-dives"]):::rest classDef done fill:#23204c,stroke:#d9b36c,color:#eceaf8,font-size:19px classDef today fill:#ab7d22,stroke:#d9b36c,color:#14132b,font-size:19px classDef rest fill:#23204c,stroke:#8f8cb8,color:#b9b7d6,font-size:19px
You have found the papers and read them. Today you write — and someone else sets the rules.
The first session where the answer is not "it depends on your judgement" but "it depends on your target journal's own page".
Learning Objectives
Objectives 4–6 next · full verbatim wording in the reference deck and curriculum
Learning Objectives
Full verbatim wording in the reference deck and curriculum
Section 01
Every policy in this session turns on one question: did the tool improve how your ideas are expressed, or did it supply the ideas?
Literacy Foundation
The spectrum is continuous; the policies are not. Most publishers draw their line somewhere in the middle — and they do not all draw it in the same place.
The Rules, In Their Words
"We distinguish various uses for AI and related technologies as: assistive (and no longer requiring disclosure), generative (requiring disclosure), and prohibitive."
Sage, Artificial intelligence policy [21]Sage lists translation and literature-review assistance as disclosable — many researchers assume they are not [21]
The Evidence — 1 of 2
Survey of 908 researchers across eight nationalities [12] · 38.1% of non-native speakers had a paper rejected over English, against 14.4% of native speakers · around 30% of early-career non-native speakers often skip English-language conferences
The Evidence — 2 of 2
Section 02
Eight rulebooks, one universal rule, and a table you should not trust for longer than a year.
The Rules, In Their Words
"Chatbots (such as ChatGPT) should not be listed as authors because they cannot be responsible for the accuracy, integrity, and originality of the work."
ICMJE [14]"AI tools cannot meet the requirements for authorship as they cannot take responsibility for the submitted work."
COPE position statement [16]Science: "artificial intelligence tools cannot be authors" [22] — and the corollary matters more: a human is accountable for every sentence, including the ones a tool wrote [14]
Series Artefact
Every row quoted from that publisher's own current policy page, fetched 2026-07-28 [17] [18] [19] · full verbatim rows in the reference deck · row order is not a ranking
Series Artefact
Fetched 2026-07-28 from each publisher's own page [20] [21] [22] [23] · category labels are ours, not the publishers' — except Sage's three tiers
Series Artefact
Fetched 2026-07-28 [14] [24] [25] · the handout carries all eight rows in one table, each with a link to the body's own page
Critical Literacy
The modal 2026 rule is not "don't" — it is "you may, and you must say so". These pages change: Science relaxed in Nov 2023 [23] [22], Elsevier renamed its declaration in June 2026 [17]. Date every policy claim — these are true as of 28 July 2026.
Critical Literacy
flowchart LR A["publisher policy page
the shape of the requirement"]:::n --> B["the journal's own guide for authors
the wording"]:::hot --> C["check it
the week you submit"]:::n classDef n fill:#23204c,stroke:#8f8cb8,color:#eceaf8,font-size:20px classDef hot fill:#23204c,stroke:#d9b36c,color:#d9b36c,font-size:20px
Heterogeneity "persist[ed] in some instances among affiliated publishers and journals" in the first systematic audit (screened 2023) [6]
Section 03
Most journals now ask. Almost nobody answers. The size of that gap has finally been measured.
The Evidence
Three figures, three windows, one study — consistent, not contradictory [1] · 70% of journals had adopted an AI policy, "primarily requiring disclosure" — and use rose "with no significant difference between journals with or without policies"
Critical Literacy
57 ÷ 0.43 ≈ 130 — two orders of magnitude of gap either way [29] [1] · group-level estimates "cannot distinguish whether AI is used merely for linguistic polishing… or for substantive text generation" [1]
The Evidence — Self-Reported Survey Data
Self-reported, publisher-run poll (Nature, March 2025); the article itself flags response bias [7] · small wonder: "improve the language" is exempt at Springer Nature and Wiley, but disclosable at Sage once it changes structure [18] [19] [21]
Section 04
If you take one operational fact from this session, take this one: AI-text detectors are not evidence, and they are not neutral.
The Evidence — 1 of 3
"the available detection tools are neither accurate nor reliable and have a main bias towards classifying the output as human-written rather than detecting AI-generated text"
Weber-Wulff et al. (2023), 12 public tools plus Turnitin and PlagiarismCheck [8] — the failure is two-sided: they miss AI text, and they misfire on human textThat was 2023's verdict on 2023's tools; the August-2026 re-examination reads the same: "Honest AI-editing results in a higher sanction risk than humanizer-assisted evasion." Evaders can opt out; honest writers cannot [30]
The Evidence — 2 of 3
The same detectors classified US eighth-grade essays with near-perfect accuracy — in that 2023 cohort they keyed on low text perplexity, penalising "individuals with limited linguistic proficiency" [9] · the skew outlived the cohort: in the 2026 re-test, unmodified human abstracts were still flagged at 9–15%, non-STEM far above STEM [30]
The Evidence — 3 of 3
The instruction was simply to "enhance the readability and flow" while preserving "original meaning and intent" — and polishing "revealed alterations in meaning for on average 2 to 3 sentences per letter": the reason Demo A ends with a diff [10]
Practice
flowchart LR A["1 · Disclose
in the required format —
suspicion becomes documented fact"]:::n --> B["2 · Keep the record
tool, version, dates,
what you asked it to do"]:::n --> C["3 · Diff every polished paragraph
against your original,
before accepting it"]:::n --> D["4 · If accused
correct and clear content,
not which tool"]:::hot classDef n fill:#23204c,stroke:#8f8cb8,color:#eceaf8,font-size:19px classDef hot fill:#1c1a3f,stroke:#d9b36c,color:#d9b36c,font-size:19px
"The main question… is whether the content is correct and clear, not which tool was used to polish the language." [11] · the record elements are the ones JAMA already requires [25] — and meaning drift is a property of rewriting, not of one model version [17] [24] [10]
Section 05
When you review a manuscript or a grant, the rules stop being about disclosure and start being about confidentiality.
The Rules, In Their Words
ICMJE gives the reason: manuscripts are "privileged communications that are authors' private, confidential property" [15]
The Rules, In Their Words
"NIH prohibits NIH scientific peer reviewers from using natural language processors, large language models, or other generative Artificial Intelligence (AI) technologies for analyzing and formulating peer review critiques…"
NIH Notice NOT-OD-23-149, released 23 June 2023 [27] — uploading application content to online generative AI tools "violates" confidentiality and integrity requirementsNSF: reviewers "prohibited from uploading any content from proposals" — while proposers are "encouraged to indicate in the project description" any generative-AI use
NSF notice to the research community [28]Reviewers now certify compliance in a modified confidentiality agreement; accessibility technologies may be excepted with prior notice [27]
Demo Preview
Demo A runs this against a rewrite prompt on the same rough paragraph and diffs both · Demo B writes the disclosure statement Elsevier requires for it [17]
References
References