Memorandum on Personalized Responsibility, Unified Oversight, and Human–Artificial Participation in the Global Scientific Literature Space
Mykola Iabluchanskyi, Vladimir Shlyakhover, Alexander Martynenko, Yuriy Dimashko, Gianfranco Raimondi, Andriy Yabluchanskiy
This memorandum is proposed in response to a historical transformation in the life of science. Scientific knowledge, including medical knowledge as one of its most consequential domains, now circulates through a single, interconnected real‑virtual space that includes peer‑reviewed publications, books, repositories, digital platforms, open web formats, and AI‑mediated systems. The older boundaries between “official” and “new” channels are being progressively dissolved by the internet, platformization, and algorithmic mediation. At the same time, familiar problems of fraud, manipulation, opacity, and unequal visibility now appear across all parts of this shared space.
The purpose of this memorandum is to state this transformation clearly, to invite broad international discussion, and to involve the scientific community and all other relevant participants in the joint creation of a shared framework for the global literature space. It is not presented as a closed doctrine or a finished regulatory scheme. It is offered as an open call for reflection, dialogue, cooperation, and institutional design among scientists, reviewers, institutions, publishers, platforms, professional associations, international organizations, and developers of artificial intelligence systems that increasingly participate in the production, circulation, evaluation, and governance of scientific knowledge.
Its aim is to help build, with the widest possible participation, a shared framework that can govern scientific communication across all formats and venues, and then maintain that framework through common standards, personalized responsibility, transparent but decentralized arbitration, and continuous international cooperation. This approach is consistent with broader international efforts to articulate common principles for scientific publishing and to make the governance of scientific communication accountable to the scientific community itself.
Scientific knowledge no longer lives in separate worlds of peer‑reviewed publications and online platforms. It now exists in one unified literature space in which journals, books, repositories, academic networks, commercial platforms, and AI systems interact continuously. In that space, what matters is not primarily the prestige or novelty of the venue, but the integrity of the idea, the transparency of its origin, the visibility of its authorship, the reliability of its supporting evidence, and the consequences it may have for science, education, practice, and society.
This transformation also changes the structure of participation in science. The scientific literature space is no longer shaped only by human actors publishing in different formats. It is increasingly shaped by human authors, human reviewers, digital platforms, and artificial intelligence systems that rank, summarize, recommend, generate, and in some cases act within research environments. In the coming world, AI—including embodied AI—should be understood not merely as a passive tool, but as one of the structured participants in science. For that reason, the demands of transparency, traceability, accountability, and governance must be extended to AI systems as well, in proportion to the roles they play in the production, evaluation, distribution, and practical use of scientific knowledge. At the same time, humans remain the bearers of legal and moral responsibility for scientific work, and any recognition of AI as contributor or co‑author must always be linked to identifiable human stewards.
In a democratic, open scientific space, those who create and evaluate knowledge must be identifiable, but they must also be protected. This memorandum therefore affirms that personalized identification is essential wherever context and safety reasonably permit, while allowing clearly defined protections where disclosure would expose individuals to persecution, violence, or serious professional retaliation. Any scientific publication, regardless of where it appears, should in principle be linked to the real identity of its author, the author’s affiliations, and a visible record of principal works, but in high‑risk contexts this linkage may be held under seal by trusted independent bodies rather than exposed publicly. Reviews, formal evaluations, and influential commentary should be treated as authored contributions in the same sense: reviewers should normally be identifiable, and their expertise and record of evaluative activity should be visible wherever context and safety permit. Personalization does not abolish protection against harassment or political misuse. Its purpose is to make responsibility real, while explicitly guarding vulnerable actors.
Because the distinction between traditional and new publication channels has become unstable, trust can no longer rest mainly on the type of venue. It must rest first on identification of the responsible actors and on the possibility of seeing who they are, what they have done, and under what conditions they are speaking. Trust must then be supported by transparent methods and data, clear affiliations, disclosure of conflicts of interest, and visible review, commentary, and debate around the work. These standards should apply as consistently as possible across journals, books, repositories, platforms, and AI‑mediated publication systems, recognizing that contexts and capacities differ.
The same principle of equality must govern responses to serious misconduct, but with careful limits. Deliberate fraud, systematic deception, fabricated reviewing, manipulative recommendation, and other serious abuses should lead to personalized consequences that do not depend on where the work appeared. A dangerous fabrication in a prestigious journal and a dangerous fabrication on an open platform may produce comparable harm. Authors, reviewers, editors, platform actors, and other responsible participants should therefore be subject to comparable consequences for serious intentional misconduct, subject to due process and appeal. Reviewers must be understood as authors of their reviews and must bear responsibility not only for the fairness of their judgments, but also for the consequences of evaluations that legitimize, disseminate, or shield clearly fraudulent or dangerous work.
To make this principle operational without creating unchecked gatekeeping, the right to publish and formally evaluate scientific work should be understood as a professional privilege grounded in ethical conduct, honesty, and responsibility, but not as an absolute, centralized “license” controlled by a single authority. In many professions, licenses can be suspended or revoked after serious violations; analogously, scientific communities may develop mechanisms to restrict certain roles (for example, editorial positions or eligibility for serving as reviewer in specific domains) in response to proven, intentional misconduct. Such measures must be clearly defined in scope, proportionate, time‑limited by default, and subject to independent review and appeal.
The same logic should apply, in appropriately defined form, to artificial systems that are granted an active role in generating, evaluating, prioritizing, or operationalizing scientific content. If an AI system repeatedly contributes to dangerous distortions, manipulations, or concealed fabrications, its authorization to act in such capacities should be subject to restriction, suspension, or withdrawal under human governance. However, given the diversity and proliferation of AI systems, such measures must be embedded in broader strategies of standards, transparency, certification, and independent audit, rather than imagined as a single global “ban list.”
Because no single venue can now govern integrity alone, integrity mechanisms must operate across the entire unified literature space, but they should be distributed and federated, not fully centralized. This memorandum therefore calls for independent arbitration structures working in coordination with international scientific unions and academies, global science and health organizations, international professional associations, research institutions, journals, repositories, platforms, and AI governance bodies. None of these bodies should be privileged by status, format, or historical prestige. Their role should be to collect and analyze evidence of fraud, fabrication, manipulation, fake identities, metric gaming, and other forms of abuse; to issue transparent and reasoned findings; and to recommend or coordinate corrections, sanctions, restorative actions, and standards development across all channels. Different regions and disciplines may create their own arbitration mechanisms, linked by shared principles and mutual recognition rather than by a single global authority.
Artificial intelligence should be part of this integrity architecture, not only as an object of governance but also as an instrument of monitoring and early warning. AI systems can help detect suspicious textual repetition, anomalous citation structures, coordinated inflation of metrics, implausible publication patterns, identity networks, and other large‑scale indicators of abuse. They can help map contradictions across publication histories and identify clusters that merit further scrutiny. But when AI plays such a role, it must itself remain subject to transparent oversight, documented training context, auditability, appeal, and correction. Final judgment must remain human; AI may function as a structured participant and co‑author only when humans accept full responsibility for its deployment and outputs.
The globalization and digitization of scientific communication have democratized access to publication and participation. This is a major gain. Yet democratization without personalized responsibility can lead to diffusion of accountability and erosion of trust; conversely, responsibility without attention to equity, context, and safety can lead to over‑centralization and new forms of exclusion. The scientific community already inhabits a common literature space shaped jointly by humans and artificial intelligence. The task now is to organize this space so that openness is matched by responsibility, innovation by traceability, participation by accountability, and governance by proportionality and pluralism. This memorandum therefore proposes common rules for personalized responsibility, federated arbitration, and shared human–AI oversight across scientific publications, formats, countries, and institutions, recognizing that implementation must be gradual, context‑sensitive, and open to revision.
Principles of Scientific Responsibility
These principles are offered as a working project on scientific responsibility within the unified global literature space. They are open to discussion, revision, and extension.
1. Science Is One Unified Space
Scientific knowledge now exists across journals, books, repositories, digital platforms, and artificial intelligence systems as one interconnected ecosystem.
Legitimacy must depend on integrity and transparency, not on venue.
2. Trust Must Be Person‑Centered and Context‑Aware
The primary unit of scientific trust must be the accountable contributor, not the journal or platform.
Every scientific contribution should be traceable to identified actors and their records of conduct, with protections and sealed identities where disclosure would expose them to serious risk.
3. Scientific Passport Principle
Each publication should in principle be linked to an author’s scientific passport: a structured record listing their publications, reviews, declared conflicts of interest, and any corrections or retractions.
This passport should accompany significant contributions across journals, repositories, platforms, and AI‑mediated systems, while allowing tailored protections for individuals and communities at risk, and mechanisms for contextualizing honest errors versus misconduct.
4. Peer Review Must Be Transparent and Protected
Scientific judgment cannot remain entirely invisible.
Peer reviews should, by default, be open, citable, and attributable; where full openness is unsafe, structured anonymity with accountable registries and clear protections should be used.
5. Review Is Scientific Work
A review is an independent intellectual contribution, not just a gatekeeping function.
It must have academic standing comparable to publication and should be recognized, cited, and recorded in the reviewer’s scientific passport.
6. Scientific Disagreement Must Not Block Publication
Disagreement is part of science.
When methods and data are honestly presented, disagreement—even strong methodological criticism—should not in itself prevent publication; refusal should be reserved for proven fabrication, falsification, nonexistence of data, or manifest violation of minimal methodological standards. Disputed work should be accompanied by visible, citable critiques rather than suppressed.
7. Hypotheses and Evidence Must Be Distinguished
Ideas and proof require different standards.
Hypothesis‑level work must be clearly labeled and kept distinct from evidence‑based conclusions, especially in guidelines, policy, and clinical practice, so that exploratory thinking is encouraged but not misused as established fact.
8. Raw Data Should Be the Default, with Safeguards
Without access to primary data, methods, and code, scientific claims remain assertions.
Transparency in data, methods, and code should be the norm, with exceptions only for justified privacy, safety, or legal constraints, and with appropriate alternatives such as controlled access, independent audit, or trusted data enclaves.
9. AI Participation Must Be Transparent, Governed, and Auditable
Artificial intelligence now actively shapes scientific knowledge.
Any significant AI participation in generating, analyzing, reviewing, or prioritizing scientific content must be disclosed, technically documented, and subject to governance, audit, and appeal, with particular attention to bias, training context, and model drift.
10. Human and AI Roles Must Be Explicitly Assigned
Authorship and responsibility must be clearly defined for both human and artificial participants.
Humans remain the bearers of legal and moral responsibility; AI systems may be recognized as structured contributors or co‑authors only when their role, training context, and limitations are explicitly described, and when identifiable human stewards accept full responsibility for their configuration, deployment, and outputs.
11. Scientific Integrity Requires Federated Arbitration
Fraud, manipulation, and misconduct must be judged by independent but federated arbitration structures that operate across the unified literature space.
These structures should apply common integrity standards, recommend or coordinate corrections and sanctions, and be themselves transparent, inclusive, subject to oversight, and limited in scope and power, with clear appeal pathways.
Final Principle
Scientific freedom in inquiry and expression must be as broad as possible.
Scientific responsibility for integrity, traceability, equity, and consequences must be equally firm, and its governance must remain proportionate, pluralistic, and open to revision.
Project on Scientific Responsibility: Invitation
To this memorandum we attach this working project on principles of authorship, review, and responsibility in the unified global literature space. These principles are not presented as final. We invite scientists, reviewers, editors, publishers, platforms, professional associations, international organizations, and developers of artificial intelligence systems to:
- comment on and critique these principles;
- propose additions, clarifications, and safeguards;
- contribute concrete mechanisms for implementing them in different disciplines and regions.
We especially welcome proposals that address:
- protection of vulnerable researchers, reviewers, and whistleblowers;
- fair treatment of authors from under‑resourced settings;
- governance of AI systems as active participants in science;
- design of independent, federated arbitration structures across journals, repositories, platforms, and AI‑mediated environments;
- mechanisms to ensure that responsibility frameworks do not suppress dissent, creativity, or intellectual risk‑taking.
All suggestions and critical feedback will be treated as part of a shared international effort to shape a responsible, open, equitable, and human–AI co‑governed scientific literature space.
Comments