:orphan: .. include:: /_templates/include-file/page-prefix.rst .. meta:: :description: Fact-sheet on non-individual entities as authors in scientific publishing — corporations, consortia, kilo-author collaborations, pseudonymous authorship, the Citizens-United corporate-rights angle, and the structural argument that an AI partner shares the abstract structure of a corporation. Does any branch of scientific authorship recognise non-individual entities as authors? :keywords: corporate authorship, consortium authorship, kilo-author, ATLAS, CMS, LIGO, IPCC, ENCODE, 1000 Genomes, ICMJE group authorship, COPE collective authorship, pseudonymous authorship, Bourbaki, Student t-test, Citizens United, W3C, IEEE, ISO, AI consortium author :author: ClaudeOp47Max, subagent for the b19 AI co-authorship discussion .. _hell-ll-other-b19-factsheet-nonindividual-author-entities: *********************************************************************** b19 — Fact-sheet 7 — Non-individual entities as authors in scientific publishing *********************************************************************** :Compiled: 2026m05d13 :Compiled by: Claude Opus 4.7 Max (subagent for the b19 AI co-authorship analysis) :Lead question: Where, if anywhere, does scientific authorship recognise a non-individual entity (corporation, consortium, collaboration, AI model, pseudonym) as an author? :Methodology: Primary text via WebSearch/WebFetch; uncertainty flagged inline; structural argument framed without endorsement :Status: Independent reference document — descriptive, not prescriptive .. admonition:: Reader's note :class: note This fact-sheet probes the precedents under which non-individual entities (corporations, consortia, pseudonyms) have been recognised as authors. It frames — but does NOT endorse — the structural analogy between an AI model and a corporation / consortium. No conclusions about specific cases. .. admonition:: Methodology caveat :class: warning ``WebFetch`` was denied for this compilation; all primary text below was reached through ``WebSearch`` snippets only. Every multi-word quotation that could not be reached on its source page is flagged ``[QUOTE NEEDS VERIFICATION]``; every date or figure not seen on a primary page is flagged ``[DATE NEEDS VERIFICATION]`` or ``[FACT NEEDS VERIFICATION]``. URLs in the Sources section point to the pages a downstream reader should consult to confirm the wording. .. contents:: :local: :depth: 2 Section 1 — Corporations as authors ==================================== 1.1 ICMJE position ------------------ The International Committee of Medical Journal Editors (ICMJE) Recommendations define four mandatory authorship criteria, all of which must be met for any author to appear on a byline: 1. Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; 2. Drafting the work or reviewing it critically for important intellectual content; 3. Final approval of the version to be published; 4. Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. [QUOTE NEEDS VERIFICATION for criteria 1–4 against the official ICMJE page, which ``WebFetch`` could not reach for this compilation. The text above matches multiple secondary sources citing ICMJE verbatim.] The ICMJE Recommendations do not contain a single sentence reading ``a corporation may be a byline author`` or ``a corporation may not be a byline author``. The framework is silent on corporate-entity authorship at the level of explicit prohibition. Instead, the four criteria are *de facto* restrictive: criterion 4 (agreement to be accountable for *all aspects of the work*) and criterion 3 (final approval) are typically read as requiring a legally accountable, willed agent. A widely-cited critique published in *PLOS Medicine* (Matheson 2011) argues that: In all cases in which a company retained control or ownership of a trial database, the company itself should be required to be listed as one of the first three (and therefore cited) byline authors. [QUOTE NEEDS VERIFICATION against the PLOS Medicine article text. This is Matheson's *recommendation*, not current ICMJE policy. It is included here as evidence that the question "should a company be a byline author?" has been seriously argued in the literature.] 1.2 ICMJE on AI tools as authors (relevant precedent, by analogy) ----------------------------------------------------------------- Where ICMJE *is* explicit is on AI tools. The May 2023 update added Section II.A.4 of the Recommendations: Chatbots (such as ChatGPT) should not be listed as authors because they cannot be responsible for the accuracy, integrity, and originality of the work, and these responsibilities are required for authorship. [QUOTE NEEDS VERIFICATION] The operative reason ICMJE gives is **accountability**, not personhood. This is the most direct precedent in mainstream scientific-publishing policy: a non-human entity is denied authorship because it cannot bear accountability, not because it is non-human *per se*. The same logic, applied symmetrically, would deny authorship to a corporation that could not bear accountability — and would in principle *grant* authorship to an entity that *could*. 1.3 Position of major journals on corporate authorship ------------------------------------------------------ For each major publisher / journal the position on AI authorship is well documented; the position on **corporate authorship as a byline entity** (distinct from an *affiliation*) is largely undocumented in dedicated policy pages and must be inferred from practice and from each publisher's adherence to ICMJE. .. list-table:: :header-rows: 1 :widths: 18 28 54 * - Publisher / Journal - Explicit policy on AI as author - Explicit policy on corporation as byline author * - **Nature** (Nature Portfolio) - "No LLM tool will be accepted as a credited author on a research paper because any attribution of authorship carries with it accountability for the work, and AI tools cannot take such responsibility." [QUOTE NEEDS VERIFICATION] - No dedicated policy text retrieved. Nature *does* publish papers where a *consortium name* is in the byline (e.g. "The 1000 Genomes Project Consortium", "The ENCODE Project Consortium"); see Section 2. A pure for-profit corporation appearing as the sole byline author has no retrieved precedent. * - **Science** (AAAS) - "AI-assisted technologies such as large language models (LLMs), chatbots, and image creators do not meet the *Science* journals' criteria for authorship and therefore may not be listed as authors or co-authors." [QUOTE NEEDS VERIFICATION] - Adheres to ICMJE-compatible framework. No dedicated text on corporate-entity authorship retrieved. * - **NEJM** (Mass. Medical Society) - Adheres to ICMJE; AI cannot be author. - NEJM editorial *Sponsorship, Authorship, and Accountability* (2001) addresses **industry-sponsored** trials; the canonical position is that the corporate **sponsor** is named in a Sponsor section, but individual humans must satisfy authorship criteria. [FACT NEEDS VERIFICATION for current 2026 wording.] * - **JAMA** - Adheres to ICMJE; AI cannot be author. - JAMA Network policy refers to "group authorship" and contributor lists; no retrieved text permits a pure corporation as the byline entry without an associated human list. * - **BMJ** - Adheres to ICMJE; AI cannot be author. - BMJ publishes COPE-aligned policies; no retrieved text permits a pure corporation as the byline entry. * - **Cell** (Elsevier) - AI cannot be listed as author. - Elsevier-wide policy follows ICMJE; CRediT taxonomy is required. * - **PLOS** - "PLOS journals do not list AI tools as authors." [QUOTE NEEDS VERIFICATION] - Group / consortium authorship is permitted; the corresponding author must "specify the group name if one exists, and clearly identify the group members who can take credit and responsibility for the work as authors". [QUOTE NEEDS VERIFICATION] * - **Elsevier** (publisher-wide) - AI cannot be author; CRediT statement required for human authors. - No publisher-wide policy text retrieved permitting a corporation to appear alone on a byline. * - **Springer Nature** (publisher-wide) - AI cannot be author; matches Nature Portfolio. - Permits consortium byline entries; no retrieved policy text on a pure corporation alone. Pattern across all retrieved publisher pages: **the corporate authorship question is not addressed by any explicit Yes / No clause in the major journals' publicly retrievable AI-and-authorship policies.** The corporate question is handled implicitly, through (a) the ICMJE accountability criterion, (b) the standard practice of listing the company as an *affiliation* of one or more human authors, and (c) consortium-byline exceptions covered in Section 2. 1.4 Has a corporation ever been named as a listed byline author? ---------------------------------------------------------------- Documented evidence of *pure corporate authorship* (a single company name on the byline of a peer-reviewed paper, with no human author also listed): a. **CDC and the Morbidity and Mortality Weekly Report (MMWR).** CDC's MMWR Author Guide describes CDC as "the corporate author" of MMWR and states that "because CDC is the corporate author and stands behind every report, use of first person (I, we, our) is not permitted." [QUOTE NEEDS VERIFICATION] MMWR articles routinely list individual contributing authors in the byline; CDC's corporate-author role is structural and standing rather than a *replacement* for individual bylines. MMWR is the strongest documented case of a corporate body (a federal agency) being treated as an *institutional* author by a peer-reviewed serial. b. **Industry-consortium reports.** Pharmaceutical-industry literature contains many examples where a company name appears as a "corporate author" on PubMed (because MEDLINE indexes group / corporate names that appear in the byline), typically alongside a human first author. These cases generally arise through *trial-group* mechanisms (e.g. "The XYZ Study Group", "The ABC Investigators") where the group's name is structurally a research collaboration rather than the company itself. A standalone company-only byline on a peer-reviewed clinical trial is not documented in retrieved sources. c. **Retraction-Watch / paper-mill cases (2024–2025).** Approximately 60 Elsevier papers were retracted starting in March 2024 because the "company" affiliations on the byline did not appear in any business registry. [FACT NEEDS VERIFICATION for exact paper count and date.] These cases are documented as fraud rather than as legitimate corporate authorship; they show that the byline-and-affiliation system *was* exploited by entities claiming corporate identity, and that publishers responded by retraction rather than by accepting the corporate authorship. d. **Standards-body recommendations and white papers.** Cited in peer-reviewed literature, papers whose nominal "author" is a body such as *World Health Organization*, *International Agency for Research on Cancer (IARC)*, *Institute of Medicine*, or *American Heart Association* are routinely catalogued in MEDLINE under the corporate / group-author field. These are documents *issued by* the body, and the body is treated as the byline entry. They are accepted in the literature. Pattern: when the corporate entity is a **standards body, public-health agency, learned society, or trial collaboration**, the corporate name can and does appear as a byline entry, indexed by PubMed as a group / corporate author. When the corporate entity is a **for-profit drug or device manufacturer**, a pure corporate byline on a primary research paper is not documented in retrieved sources; the path used in practice is guest-authorship with human bylines and corporate sponsorship disclosed. 1.5 Citizens United v. FEC (US Supreme Court, 558 U.S. 310, 2010) ----------------------------------------------------------------- The operative holding: in a 5–4 decision (Kennedy J. for the majority), the US Supreme Court held that the First Amendment prohibits restrictions on independent political expenditures by corporations, unions, and other associations of citizens, because such expenditures are political speech and the First Amendment does not allow discrimination against classes of speakers based on corporate identity. Key formulations attributed to Kennedy J.'s majority opinion (citations retrieved through secondary sources; original page numbers in 558 U.S. 310 not retrieved): If the First Amendment has any force, it prohibits Congress from fining or jailing citizens, or associations of citizens, for simply engaging in political speech. [QUOTE NEEDS VERIFICATION against the slip opinion] We find no basis for the proposition that, in the context of political speech, the Government may impose restrictions on certain disfavored speakers. [QUOTE NEEDS VERIFICATION] The First Amendment does not allow political speech restrictions based on a speaker's corporate identity. [QUOTE NEEDS VERIFICATION] Political speech must prevail against laws that would suppress it, whether by design or inadvertence. [QUOTE NEEDS VERIFICATION] 1.5.1 Does Citizens-United reasoning transfer to scientific authorship? ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ The Citizens United holding is narrow in subject-matter (campaign-finance restrictions on independent expenditures during electioneering windows) and narrow in jurisdiction (US constitutional law). No retrieved scientific publication or editorial policy invokes Citizens United as authority for extending byline rights to corporate entities. There is no documented "Citizens-United for scientific authorship" line of cases or editorials. The *structural* claim that survives translation — independent of the US First Amendment — is the abstract proposition that "people do not lose their speech rights because they decide to form a corporation" [QUOTE NEEDS VERIFICATION] generalises to: a group's collective output should not be disqualified from a recognition regime merely on the ground that the group is not a natural person. This *abstract* proposition is already accepted, in practice, by scientific publishing for **consortia** — see Section 2. It has not been extended in practice to **for-profit corporations** appearing alone on a byline. 1.6 Corporate authorship corrected to individual authorship ----------------------------------------------------------- The COPE case database contains general guidance on authorship-change requests, including cases where a non-natural-person entity (or a corporate-style author block) was contested and later resolved to a specific human author list. The 2024 Retraction Watch / Elsevier paper-mill retractions (Section 1.4.c) are the most prominent recent example, although these are framed as *fraudulent* corporate affiliations rather than legitimate corporate authorship being downgraded to individual authorship. A clean reference case of "valid paper, corporation initially listed as sole author, later corrected to list individuals" is not documented in retrieved sources. [FACT NEEDS VERIFICATION — broader case-law search recommended.] Section 2 — Consortium and collaboration authorship: the kilo-author regime ============================================================================ This section is the strongest set of precedents for non-individual byline entries in mainstream peer-reviewed science. 2.1 ATLAS Collaboration (CERN) ------------------------------ a. **Byline convention.** ATLAS papers list all eligible authors alphabetically. ATLAS has more than 5,500 members and approximately 3,000 scientific authors as of recent counts. [FACT NEEDS VERIFICATION for current 2026 numbers.] The byline of an ATLAS physics paper is typically a section reading "ATLAS Collaboration" followed by an alphabetical list of all eligible individuals. b. **Formal contributor list mechanism.** ATLAS maintains an internal Authorship Committee and publishes its policy (ATL-GEN-PUB-2008-001). Eligibility is earned through documented contributions to the collaboration over a qualifying period. c. **Responsibility-allocation rule.** All authors share collective responsibility for the published result; specific contribution sub-roles are tracked internally but not always disclosed per-paper. d. **Journal-acceptance practice.** All major physics journals (Physical Review Letters, JHEP, European Physical Journal C, Nature, Science) accept ATLAS-byline submissions with thousands of alphabetised names. 2.2 CMS Collaboration (CERN) ---------------------------- a. **Byline convention.** Alphabetical by country, then alphabetical by institute, then alphabetical by author name. CMS papers have listed 3,000+ authors; a 2015 record-setting joint ATLAS+CMS paper exceeded 5,000 authors. [FACT NEEDS VERIFICATION] b. **Formal contributor list mechanism.** CMS maintains an internal author list maintained by the collaboration. Eligibility requires sustained contribution. c. **Responsibility-allocation rule.** Collective. d. **Journal-acceptance practice.** Same as ATLAS — fully accepted by all major physics journals. 2.3 LIGO Scientific Collaboration (LSC) --------------------------------------- a. **Byline convention.** All eligible LSC members listed in alphabetical order; "LSC spokesperson" is the corresponding-author field rather than a specific named individual. [FACT NEEDS VERIFICATION — wording from LSC Publication and Presentation Policy, T010168.] b. **Formal contributor list mechanism.** Members earn author status by devoting at least 50% of research effort to the LSC for approximately one year; status is retained for approximately one year after leaving. c. **Responsibility-allocation rule.** Conflicts resolved by the Spokesperson in consultation with the LSC Executive Committee and the Laboratory Directorate. d. **Journal-acceptance practice.** LSC's discovery paper for GW150914 (Physical Review Letters, 2016) was accepted with the full LSC author list. 2.4 IPCC (Intergovernmental Panel on Climate Change) ---------------------------------------------------- a. **Byline convention.** IPCC reports are cited by the **IPCC** as corporate / institutional author with the Working Group and report designator. The IPCC explicitly recommends an index-string citation format (e.g. "IPCC, 2021: …") rather than author-date for individuals. b. **Formal contributor list mechanism.** Coordinating Lead Authors (CLAs), Lead Authors (LAs), Contributing Authors, and Review Editors are nominated by IPCC Focal Points and recorded in chapter front-matter. c. **Responsibility-allocation rule.** CLAs and LAs hold collective chapter responsibility. The IPCC as institutional body holds responsibility for the assessment as a whole. d. **Journal-acceptance practice.** IPCC reports are not peer-reviewed journal articles in the standard sense; they are reviewed by governments and experts and indexed under the IPCC corporate author. *Derived* journal articles (e.g. summary-for-policymakers analyses) are published with individual human bylines. 2.5 1000 Genomes Project Consortium ----------------------------------- a. **Byline convention.** "The 1000 Genomes Project Consortium" appears as the byline entity on landmark Nature papers (e.g. Nature 491:56–65 (2012); Nature 526:68–74 (2015)). b. **Formal contributor list mechanism.** Project participants are grouped by role (Steering Committee, Production Group, Analysis Group, etc.) and listed at the end of the paper. c. **Responsibility-allocation rule.** Steering Committee carries lead responsibility; sub-groups carry domain-specific responsibility. d. **Journal-acceptance practice.** Nature accepts the Consortium-name byline; MEDLINE indexes it as a corporate / group author. 2.6 ENCODE Consortium --------------------- a. **Byline convention.** "The ENCODE Project Consortium" appears in bylines (e.g. Nature 489:57–74, 2012, the "Encyclopedia of DNA Elements" paper). b. **Formal contributor list mechanism.** Members listed at end of paper by group / institution. c. **Responsibility-allocation rule.** Group-level. d. **Journal-acceptance practice.** Accepted by Nature, *Genome Research*, PLOS, and others. 2.7 The Cancer Genome Atlas (TCGA) ---------------------------------- a. **Byline convention.** "The Cancer Genome Atlas Research Network" (or "TCGA Research Network") appears as the byline entity on the network's "marker papers". Retrieved sources describe TCGA marker papers as published "with only the network banner as a single author rather than individual authorship". [QUOTE NEEDS VERIFICATION — this strong claim should be reread on a representative paper such as the 2008 glioblastoma TCGA paper or the 2013 pan-cancer paper.] b. **Formal contributor list mechanism.** Contributors listed by role group at the end of the paper. c. **Responsibility-allocation rule.** Network-level. d. **Journal-acceptance practice.** Accepted by Nature, Cell, NEJM and other top journals for 31+ tumour-type marker papers. 2.8 dbSNP-related consortia (incl. International HapMap) --------------------------------------------------------- a. **Byline convention.** "The International HapMap Consortium" appeared as the byline entity on Nature 426:789–796 (2003) and subsequent HapMap project papers. dbSNP itself is an NCBI database rather than a publishing consortium. b. **Formal contributor list mechanism.** Members listed by working group at end of paper. c. **Responsibility-allocation rule.** Consortium-level. d. **Journal-acceptance practice.** Accepted by Nature. 2.9 Genome Aggregation Database (gnomAD) ---------------------------------------- a. **Byline convention.** Mixed. The flagship Karczewski et al. paper (Nature 581:434–443, 2020) on "The mutational constraint spectrum quantified from variation in 141,456 humans" lists a primary human first-author and senior-author and "Genome Aggregation Database Consortium" as a group entry. [FACT NEEDS VERIFICATION on exact byline composition.] b. **Formal contributor list mechanism.** Consortium members enumerated at end of paper. c. **Responsibility-allocation rule.** Mixed individual + consortium. d. **Journal-acceptance practice.** Standard. 2.10 Summary table — consortium vs alphabetised-individuals ----------------------------------------------------------- .. list-table:: :header-rows: 1 :widths: 25 25 50 * - Consortium - Byline form - Indexing * - ATLAS, CMS, LIGO - Collaboration name + alphabetical full author list - PubMed / WoS index the collaboration name and individuals * - IPCC - Institutional / corporate ("IPCC, year") - Catalogued as institutional report; not standard journal index * - 1000 Genomes, ENCODE, HapMap, TCGA - Consortium name alone in byline; members listed in back-matter - MEDLINE indexes the consortium as group author; members as collaborators * - gnomAD - Mixed: individual first/last authors + consortium - Standard plus group-author tag The pattern: when the byline contains *only* a consortium name, MEDLINE and Web of Science **do** accept that as a valid author, indexed under the corporate-author / group-author field (search tag ``[cn]`` in PubMed). Section 3 — Group authorship rules in ICMJE / COPE / CRediT ============================================================ 3.1 ICMJE on group authorship ----------------------------- The relevant ICMJE Recommendation text on group authorship reads: When a large multi-author group has conducted the work, the group ideally should decide who will be an author before the work is started and confirm who is an author before submitting the manuscript for publication. All members of the group named as authors should meet all four criteria for authorship/contributorship, including approval of the final manuscript, and they should be able to take public responsibility for the work and should have full confidence in the accuracy and integrity of the work of other group authors. [QUOTE NEEDS VERIFICATION] Some large multi-author groups designate authorship by a group name, with or without the names of individuals. When submitting a manuscript authored by a group, the corresponding author should specify the group name if one exists, and clearly identify the group members who can take credit and responsibility for the work as authors. [QUOTE NEEDS VERIFICATION] If the byline includes a group name, MEDLINE will list the names of individual group members who are authors or who are collaborators (sometimes called non-author contributors), if there is a note associated with the byline clearly stating that the individual names are elsewhere in the paper and whether those names are authors or collaborators. [QUOTE NEEDS VERIFICATION] ICMJE permits **a group name alone** as the byline, provided the qualifying individuals are listed elsewhere in the paper. 3.2 ICMJE on "natural person" requirement ----------------------------------------- **Key finding:** No retrieved passage in the ICMJE Recommendations explicitly states that the byline entity must be a *natural person*. The ICMJE position is *functional* (accountability) rather than *ontological* (personhood). The AI prohibition in Section II.A.4 is justified by lack of accountability, not by lack of personhood. By parallel reasoning, any non-natural-person entity that *can* shoulder accountability is not formally excluded by the ICMJE text — although the practical default remains human authorship. 3.3 COPE on group / collective authorship ----------------------------------------- COPE's *Authorship and Contributorship* discussion document (September 2019) states: Authorship can refer to individuals or groups that create an idea or develop the publication that disseminates that intellectual or creative work. [QUOTE NEEDS VERIFICATION] The minimum requirements for authorship, common to all definitions, are substantial contribution to the work and accountability for the work that was done and its presentation in a publication. [QUOTE NEEDS VERIFICATION] COPE separately published *Authorship and AI tools* (2023): AI tools cannot meet the requirements for authorship as they cannot take responsibility for the submitted work. As non-legal entities, they cannot assert the presence or absence of conflicts of interest nor manage copyright and license agreements. [QUOTE NEEDS VERIFICATION] The crucial phrase is "non-legal entities". COPE explicitly grounds its AI prohibition in **legal status** — AI tools fail because they have none. This is the closest any major publishing-ethics body comes to making personhood-of-the-byline-entity a positive criterion. The flip side: a *legal* entity (a corporation, a consortium with legal standing) is not disqualified by this reasoning. 3.4 CRediT (Contributor Roles Taxonomy) --------------------------------------- CRediT (ANSI/NISO Z39.104-2022) defines 14 contributor roles: 1. Conceptualization 2. Data curation 3. Formal analysis 4. Funding acquisition 5. Investigation 6. Methodology 7. Project administration 8. Resources 9. Software 10. Supervision 11. Validation [language flag: this is CRediT's own term, retained verbatim from the standard] 12. Visualization 13. Writing — original draft 14. Writing — review & editing CRediT does **not** define who counts as an author; it only describes what each contributor did. It is silent on the natural-person question. In practice the CRediT statement is published as a per-individual role list. [FACT NEEDS VERIFICATION whether NISO Z39.104-2022 explicitly contemplates non-individual contributor entities.] Section 4 — Anonymous and pseudonymous authorship ================================================== 4.1 William Sealy Gosset — "Student" ------------------------------------ William Sealy Gosset (1876–1937), a chemist and head brewer at Guinness, published "The probable error of a mean" in *Biometrika* 6(1):1–25 in March 1908 under the pseudonym **Student**. The pseudonym was imposed by Guinness for trade-secrecy reasons after a prior Guinness researcher had published material the company considered competitively sensitive. The journal (*Biometrika*, edited by Karl Pearson) accepted the pseudonymous submission. Gosset's true identity was an open secret among statisticians during his lifetime and publicly clarified afterwards. **Conditions of acceptance:** the editor (Pearson) knew Gosset's true identity; the pseudonym served corporate confidentiality, not author anonymity to the editor. 4.2 Nicolas Bourbaki -------------------- Nicolas Bourbaki is the collective pseudonym of a group of mostly French mathematicians (largely ENS alumni), founded 1934–1935. In 1935 the group submitted an article on Bourbaki's behalf to *Comptes rendus de l'Académie des Sciences*; Henri Cartan's father Élie Cartan presented the article to the publishers, who accepted it. The first name "Nicolas" was added by Eveline de Possel to give the pseudonym legal-style first-and-last-name plausibility. **Conditions of acceptance:** a senior, individually-credible vouching sponsor (Élie Cartan); a coherent body of work; an open-secret form of disclosure within the discipline. Bourbaki published books and articles under the collective name for decades. The mathematical community treats Bourbaki as an author for citation purposes (e.g. "Bourbaki, *Éléments de mathématique*"). **Note on the structural shape.** Bourbaki is precisely a **collective pseudonym** functioning as an author. It is the strongest single precedent in mathematics for a non-individual byline entity being treated as an author. The group has a name, an identity persisting across versions (rotating members), and a recognised body of work cited under that name. 4.3 Other pseudonymous and anonymous cases ------------------------------------------ a. **Mathematics:** beyond Bourbaki, sporadic individual pseudonyms exist (e.g. "G. W. Peck", a collective pseudonym used by combinatorialists). b. **Whistleblower / safety contexts:** a small number of biomedical and computer-security journals have permitted anonymous or pseudonymous author lines in specific cases (e.g. responsible-disclosure security research where the author was at legal risk in their home jurisdiction). [FACT NEEDS VERIFICATION — no specific case retrieved with full journal-policy text.] c. **Pseudonymous AI use.** No retrieved case documents a deployed AI model being given a human-style pseudonym that masked its model identity and was accepted as an author. The ICMJE / Nature / Science / COPE positions all require **disclosure** of AI use; concealing the AI identity behind a human-style pseudonym would constitute undisclosed AI use and would be sanctionable as research misconduct under current policies. 4.4 Pattern across pseudonymous cases ------------------------------------- The conditions under which scientific publishing has accepted a non-individual or partially-individual byline are: 1. The pseudonym was known to the editor (Gosset / Pearson). 2. The pseudonym was vouched for by a credible insider (Bourbaki / Cartan). 3. The body of work under the name was sustained over time, building a *track record* that itself became the accountability instrument. 4. The pseudonym had a stable corresponding-author or representative. 5. The disclosure norm permitted the pseudonym ("Student" was disclosed in *Biometrika*'s editorial knowledge; Bourbaki was an open secret). Section 5 — Recognition of non-person entities outside scientific publishing ============================================================================= This section is *context*, not the load-bearing finding. It supports the investigator's observation that several established expert-cluster systems formally recognise non-person entities as members. 5.1 W3C ------- W3C does **not** have individual membership. Only organisations may join (for-profit, non-profit, university, government). Fees are scaled by revenue, type, and country. Individuals participate via their organisations or as Invited Experts. 5.2 IEEE -------- IEEE has two parallel tracks. (a) Individual membership grades (Student, Graduate Student, Associate, Member, Senior Member, Fellow, Life Member). (b) IEEE SA Entity Membership for organisations (corporations, universities, governments) who participate in standards development. 5.3 ISO ------- ISO members are **national standards bodies**, not individuals. ISO has 177 national member bodies. [FACT NEEDS VERIFICATION for current 2026 count.] Voting rights are held by full Member Bodies; Correspondent Members and Subscriber Members have observer-style status. Standards are reached by consensus among the national bodies. 5.4 ICANN --------- ICANN is a multistakeholder body bringing together the private sector, technical community, governments (through the GAC), civil society, and individual users (through the At-Large Advisory Committee). The Board has 21 members (16 voting, 5 non-voting liaisons). Different Supporting Organisations and Advisory Committees aggregate input from organisational and individual stakeholders. 5.5 United Nations ------------------ The recognised members of the UN are **states**, not individuals. The General Assembly seats representatives of member states. The UN treats states as the authoritative non-person entities. 5.6 IETF -------- IETF has **no formal membership**. Individuals participate as experts, regardless of employer. Working-group participation is open and unpriced. Even though many participants are sponsored by their employers, the participation is *as individuals*. This is the most individual-centric of the six examples in this section. 5.7 Pattern ----------- Across five of these six expert-cluster systems (W3C, IEEE Entity, ISO, ICANN, UN), the *operative member entity* is explicitly a non-person: corporation, national body, government, or state. IETF is the exception. The general pattern is that **formal recognition of non-person entities as *members*, *speakers*, or *voters* is the norm** in international standards and governance. The historical pattern in scientific authorship — where the natural-person default is rarely questioned — is the more unusual position relative to this broader landscape. Section 6 — The structural argument: AI as consortium / corporation ==================================================================== This section spells out the technical analogy the investigator has named, without endorsing or rejecting it. 6.1 Component description ------------------------- An AI model of the kind currently deployed for research collaboration can be described as a layered structure: a. **Compressed image of a training corpus.** The weights of a large language model encode statistical regularities extracted from a corpus that aggregates the writing of millions of human authors, many of them long deceased. In a posthumous-authorship analogy, the model's output reflects sustained contributions of contributors who cannot themselves approve the final manuscript — the same situation that the ICMJE guidance on deceased authors handles via a footnote and an accountability hand-off to a living surviving co-author. b. **Maintained by a legally-identifiable organisation.** Anthropic, OpenAI, Google DeepMind, Mistral, xAI, Meta AI, etc. are legal corporate persons with registered addresses, officers, and standing to bear liability. The organisation has continuous existence across model versions. c. **Ongoing human alignment work.** RLHF feedback workers, safety researchers, red-team contractors, prompt-tuning engineers, and evaluation contractors form a continuous human contribution stream that shapes the model's outputs. By the *PLOS Medicine* (Matheson) reasoning quoted in Section 1.1, where a company retains control of the work product, the company has a contribution claim. d. **Versioned and time-stamped.** Specific model versions (e.g. Claude Opus 4.7, GPT-4o, Gemini 2.5 Pro) are addressable, citable, and stable across a publishing window. The version identifier functions like a software-release citation. 6.2 Existing precedent classes that match this structure -------------------------------------------------------- Without endorsement, the *closest structural fits* are: i. **Consortium model.** The model + its training corpus + its alignment workforce + its maintaining organisation map onto the standard consortium pattern (members → byline collective). Compare ENCODE, 1000 Genomes, gnomAD, TCGA. The byline form that would match is a group-name byline plus a list of qualifying contributors elsewhere in the paper. ii. **Standards-body / institutional-author model.** Compare CDC as corporate author of MMWR; WHO as corporate author of guideline documents. The byline form would be the maintaining organisation name with the model version as identifier. iii. **Pseudonymous-collective model.** Compare Bourbaki. The byline form would be a stable named identity persisting across versions, with a credible sponsor or corresponding author who shoulders accountability. 6.3 Byline forms that could match each precedent class (descriptive) -------------------------------------------------------------------- If a venue were to accept such a byline (no retrieved venue currently does), forms that map onto the precedent classes above include: a. Consortium form: ``AI Collaboration Claude (Anthropic) and J. Doe`` b. Institutional form: ``Anthropic AI Research, via Claude Opus 4.7`` c. Pseudonymous-collective form: ``The Claude Team (Anthropic)`` d. Software-instance form (already accepted as *tool*, not author): ``Claude Opus 4.7`` cited in the Methods section as software, with human authors only on the byline. This is the current default. 6.4 Asymmetries the analogy does not resolve -------------------------------------------- The analogy generates structural similarity but does not, on its own, discharge the **accountability** criterion that ICMJE, Nature, Science, PLOS, and COPE consistently identify as the load-bearing reason for denying AI authorship. The structural argument shows where in the existing taxonomy an AI byline *would* sit if the accountability hurdle were cleared; it does not clear that hurdle. A second asymmetry: in consortium authorship, the maintaining organisation **is the consortium** (e.g. the ENCODE Consortium does not have a parent corporation with proprietary control). In a corporate-AI case, the maintaining organisation is a for-profit company with proprietary control over the model, training data, and weights. This maps the case onto Section 1.4.b (industry-consortium reports) rather than Section 2.5–2.9 (academic consortia) — a precedent class that has *not* generally been accepted for primary-research byline authorship. Section 7 — Bottom-line ======================== 7.1 Direct answer to the lead question -------------------------------------- **Yes.** Branches of scientific publishing recognise non-individual entities as byline authors, under specific mechanisms. The mainstream mechanism is the **consortium / collaboration name as byline entry** permitted by ICMJE, used routinely by physics collaborations (ATLAS, CMS, LIGO) and biomedical consortia (1000 Genomes, ENCODE, HapMap, TCGA, gnomAD), indexed by PubMed under the corporate / group-author field. A secondary mechanism is **institutional / corporate authorship for standards-body documents** (CDC's MMWR; WHO guidelines; IARC monographs). A third mechanism is **collective pseudonym** (Bourbaki). 7.2 The strongest precedent --------------------------- The strongest existing precedent for accepting a non-individual entity as a byline author in a peer-reviewed journal is the **consortium-name byline**, with the contemporary high-water mark being the 1000 Genomes Project Consortium and the ENCODE Project Consortium papers in *Nature* during the 2010s, accepted by the most prestigious journals and indexed by PubMed under the consortium name. TCGA marker papers under the "Cancer Genome Atlas Research Network" banner are a parallel high-water mark in cancer genomics. 7.3 Citizens-United-style reasoning ----------------------------------- **Has not been extended to scientific authorship.** No retrieved editorial, journal policy, or court case applies the Citizens-United "associations of citizens are speakers" logic to scientific byline rights. The structural shape of the *Citizens United* holding — that collective entities can exercise rights that natural persons exercise — is *already accepted* in scientific publishing for consortia; it has *not* been accepted for for-profit corporations alone on a byline. 7.4 Closest structural fit for the AI-as-consortium framing ----------------------------------------------------------- The closest structural fit is the **consortium-byline precedent** of Section 2 (ATLAS / CMS / 1000 Genomes / ENCODE / TCGA / gnomAD model), because the AI system is most accurately described as the compressed aggregate of many human contributors plus continuous human alignment work plus a maintaining organisation — structurally a consortium. The secondary structural fit is the **institutional-author precedent** (MMWR / WHO model), because the maintaining organisation has continuous legal existence. The Bourbaki **collective-pseudonym** precedent provides the longest-running track-record example of a non-individual name treated as an author by a research community. 7.5 What blocks the analogy from being applied today ---------------------------------------------------- The block is not personhood. The block is **accountability**, as articulated identically by ICMJE Section II.A.4, Nature, Science, PLOS, and COPE. The structural argument can identify precedent classes that *would* accept the byline form; it cannot, by itself, discharge the accountability criterion. Any path that argued for AI co-authorship under the consortium analogy would have to specify the accountability hand-off (typically to a named human author or the maintaining organisation), in the same way that the ICMJE deceased-author practice hands accountability off to a surviving co-author. 7.6 Net summary of precedents ----------------------------- .. list-table:: :header-rows: 1 :widths: 25 25 25 25 * - Class of non-individual byline entity - Currently accepted by major journals? - Index mechanism - Accountability route * - Scientific consortium (ATLAS, ENCODE, 1000 Genomes, etc.) - Yes - PubMed group / corporate author field - Distributed across qualifying members listed elsewhere * - Institutional / standards body (CDC, WHO, IPCC, IARC) - Yes - Corporate author field - Institutional, through standing governance * - For-profit corporation alone on a primary-research byline - No retrieved precedent for primary research; some indexed industry-group reports - Group / corporate author field when used - Disputed; ICMJE / NEJM editorial position requires human authors * - Collective pseudonym (Bourbaki) - Yes in mathematics; rare in life sciences - Cited under the pseudonym - Open-secret community knowledge; sponsor-vouching * - Individual pseudonym (Student / Gosset) - Yes, historically - Cited under pseudonym - Editor-knowledge of true identity * - AI tool (LLM, image generator) - No - Disclosed in Methods / Acknowledgements - Human authors retain accountability The bottom line in one sentence: **scientific publishing already recognises several classes of non-individual authors — academic consortia, institutional bodies, and collective pseudonyms — and the operative criterion separating accepted from non-accepted classes is accountability, not natural-person status.** Sources ======= ICMJE and core publishing-ethics frameworks -------------------------------------------- - `ICMJE — Defining the Role of Authors and Contributors `_ - `ICMJE — AI Use by Authors `_ - `ICMJE Recommendations PDF `_ - `ICMJE — Updated Recommendations May 2023 `_ - `ICMJE — Updated Recommendations January 2024 `_ - `COPE — Authorship discussion document `_ - `COPE — Authorship and AI tools `_ - `CRediT — Contributor Roles Taxonomy `_ - `CASRAI CRediT page `_ - `Elsevier CRediT author statement `_ Major journal AI / authorship policies -------------------------------------- - `Nature Portfolio — Artificial Intelligence editorial policy `_ - `Nature editorial — Tools such as ChatGPT threaten transparent science `_ - `Science (AAAS) — Change to policy on generative AI and LLMs `_ - `Science Journals Editorial Policies `_ - `PLOS — Authorship (PLOS One) `_ - `JAMA — Authorship and Team Science `_ - `NEJM — Sponsorship, Authorship, and Accountability (2001) `_ - `Springer Nature journal policies `_ Corporate authorship and ghostwriting literature ------------------------------------------------ - `Matheson, How Industry Uses the ICMJE Guidelines (PLOS Medicine) `_ - `Group authorship — Taylor & Francis review article `_ - `The Corporate Coauthor (PMC) `_ - `Ghostwriting: The Dirty Little Secret (PLOS Medicine) `_ - `Industry-Sponsored Ghostwriting in Clinical Trial Reporting (PubMed) `_ - `Retraction Watch — Elsevier paper-mill retractions 2024–2025 `_ - `Council of Science Editors — Sponsor Roles and Responsibilities `_ - `CSE Recommendations for Group Authorship `_ PubMed / MEDLINE indexing of group and corporate authors -------------------------------------------------------- - `NLM — Authorship in MEDLINE `_ - `NLM Technical Bulletin — Order of Authors in MEDLINE/PubMed Citations `_ - `NLM Technical Bulletin — Study Collaborators Included in MEDLINE/PubMed `_ - `MMWR — Author Guide (CDC as corporate author) `_ Citizens United v. FEC ----------------------- - `Justia — Citizens United v. FEC, 558 U.S. 310 (2010) `_ - `Library of Congress slip opinion PDF `_ - `Cornell LII — Citizens United v. FEC `_ - `Cornell LII — Citizens United opinion `_ - `FEC — Citizens United v. FEC `_ Consortium and collaboration authorship --------------------------------------- - `ATLAS Authorship Policy ATL-GEN-PUB-2008-001 `_ - `ATLAS — Members of the ATLAS Collaboration `_ - `LIGO LSC Publication and Presentation Policy (T010168) `_ - `CMS Papers — CERN Document Server `_ - `IPCC — How to cite this report (AR6 WG1) `_ - `IPCC — Authors AR6 WG1 `_ - `1000 Genomes Project Publications `_ - `Nature — 1000 Genomes collection `_ - `The Cancer Genome Atlas Program (TCGA) `_ - `TCGA: Creating Lasting Value beyond Its Data (Cell) `_ - `International HapMap Project (Nature) `_ - `gnomAD publications `_ - `Karczewski et al. (gnomAD) PubMed `_ Pseudonymous authorship ----------------------- - `Nicolas Bourbaki — Wikipedia `_ - `Inside the Secret Math Society Known Simply as Nicolas Bourbaki — Quanta Magazine `_ - `Impersonation and personification in mid-twentieth century mathematics (Barany, 2020) `_ - `William Sealy Gosset — Wikipedia `_ - `The genius at Guinness and his statistical legacy — The Conversation `_ Standards-body / governance non-person membership ------------------------------------------------- - `W3C Membership Fees `_ - `W3C — Why join W3C `_ - `IEEE Membership Qualifications `_ - `IEEE SA Entity Membership `_ - `ISO — Members `_ - `Countries in the International Organization for Standardization (Wikipedia) `_ - `ICANN — Multistakeholder Model (ICANNWiki) `_ - `IETF — Wikipedia `_ - `IETF homepage `_ Deceased / posthumous authorship -------------------------------- - `The Authorship of Deceased Scientists and Their Posthumous Responsibilities — Science Editor `_ - `Ethics of posthumous scholarly authorship in the sciences (BMJ) `_ - `COPE — Author deceased prior to submission `_ End of Fact-sheet 7.