03 · What You Need to Know
Understanding the Ethical Tension Between Open Science and Responsible Research
What Does Scientific Openness Mean?
Scientific openness refers broadly to practices that make research knowledge, processes, and outputs more accessible for examination, reuse, and further development.
Depending on the field, these practices may include open-access publication, sharing datasets and analytical code, preregistration, transparent reporting, and making research materials available for independent evaluation.
The UNESCO Recommendation on Open Science recognizes the value of openness while also acknowledging legitimate restrictions related to privacy, confidentiality, security, and other protected interests.
Scientific openness should therefore not be interpreted as an unconditional requirement to distribute every research record without access controls.
The relevant objective is to improve access to knowledge and opportunities for scrutiny in ways compatible with applicable ethical and legal obligations.
What Does Responsible Research Require?
Responsible research involves conducting and communicating scholarly work with integrity, appropriate accountability, respect for relevant rights, and attention to foreseeable consequences.
These responsibilities may concern the accuracy of reported findings, protection of participants, confidentiality, appropriate attribution, and the potential consequences of research applications.
The European Code of Conduct for Research Integrity identifies principles including reliability, honesty, respect, and accountability.
Responsible research therefore includes transparency, but it also requires judgment about circumstances in which unrestricted disclosure could create unnecessary harm.
For example, protecting identifiable interview recordings may be compatible with providing a detailed explanation of the qualitative analytical process.
Are Scientific Openness and Responsible Research Opposing Principles?
Not inherently. Openness often supports responsible research by enabling independent scrutiny, error detection, and accountability.
However, particular forms of disclosure may conflict with other legitimate responsibilities.
The distinction is between openness as a general scientific commitment and unrestricted release as one possible implementation of that commitment.
Scientific openness
Supports accessibility, transparency, verification, and responsible reuse of research knowledge and materials.
Responsible research
Requires integrity, accountability, respect for relevant rights, and proportionate attention to foreseeable harm.
Ethical tension
Arises when a particular disclosure improves accessibility but creates risks that other research obligations require researchers to address.
The appropriate response should identify the specific conflict rather than assume that the two commitments are fundamentally incompatible.
When Does Openness Create a Genuine Ethical Conflict?
A genuine conflict arises when there are defensible reasons both to disclose and to restrict particular information.
For example, public release of a detailed dataset may support independent reanalysis, while the same release could expose sensitive participant information.
The conflict is not resolved merely by declaring one value more important than the other.
Researchers should identify the scientific benefit of disclosure, the foreseeable harm, the evidence supporting the concern, and whether alternative arrangements can preserve both interests.
Where the risk is speculative or minor, unrestricted sharing may remain appropriate. Where serious harm is plausible and difficult to mitigate, restrictions may be justified.
What Are Common Situations Where Openness and Responsibility Conflict?
| Research material |
Value of openness |
Potential conflict |
| Individual participant data |
Independent analysis and verification. |
Privacy, confidentiality, and reidentification risks. |
| Qualitative interview recordings |
Verification of transcription and interpretation. |
Recognizable voices and sensitive personal accounts. |
| Detailed technical procedures |
Replication and methodological evaluation. |
Transfer of capabilities that could enable harmful conduct. |
| Trained AI models |
Independent evaluation and beneficial reuse. |
Unauthorized deployment or harmful adaptation. |
| Community-level datasets |
Examination of group differences and inequalities. |
Stigmatization or misuse of sensitive collective information. |
| Restricted institutional records |
Transparency of evidence and decisions. |
Legal, contractual, or security obligations. |
These examples identify possible conflicts, not automatic reasons for withholding. The circumstances of each project determine whether a meaningful risk exists.
Does Open Science Require All Research Data to Be Public?
No. Open-science principles generally support making research materials as accessible as appropriate, but they do not eliminate privacy, confidentiality, security, or other legitimate restrictions.
Researchers should distinguish making data publicly downloadable from making them available through controlled arrangements.
For example, authorized researchers may be permitted to analyze restricted records in a secure environment without obtaining unrestricted copies.
However, controlled access can limit participation and impose administrative burdens. Researchers should explain such restrictions and avoid claiming that restricted data are openly available.
The applicable requirements depend on the funder, institution, jurisdiction, and publication venue.
Can Openness Conflict With Participant Consent?
Yes. Participants may have agreed to specific research uses without authorizing unrestricted public distribution of their information.
For example, consent to an interview does not automatically authorize publication of the original audio recording.
Researchers should examine the scope of consent, applicable privacy law, ethics approval, and any authorized alternatives for data processing.
Where additional sharing was not contemplated, further review or authorization may be necessary.
However, consent alone is not always sufficient to justify release, particularly when data contain information about other individuals or create substantial reidentification risks.
Does Anonymization Eliminate the Conflict?
Not necessarily. Removing direct identifiers can reduce disclosure risks, but remaining information may still permit reidentification.
For example, a dataset containing a distinctive combination of academic program, age, institution, and learning behavior may identify individuals in a small population.
The effectiveness of anonymization depends on the data, plausible external information, and relevant technical and organizational circumstances.
Researchers should also consider whether detailed group-level information could create collective harms even when individuals cannot be identified.
Anonymization is therefore a risk-reduction process rather than a universal guarantee of safety.
Can Open Methods Create Dual-Use Risks?
Potentially. Some research methods or technical outputs can support beneficial and harmful applications.
A model developed to identify students needing support might also be used for intrusive monitoring or discriminatory classification.
However, the mere possibility that a method could be adapted does not establish that publishing it creates a serious risk.
Researchers should examine whether disclosure materially increases the feasibility or scale of harmful use.
The principles of dual-use research provide a foundation for distinguishing plausible misuse from speculation.
Is Restricting Access the Same as Concealing Research?
No. Restricting access to particular materials can be compatible with transparent reporting when the restriction is justified and accurately described.
Concealment becomes a concern when researchers misrepresent what was done, withhold information merely to avoid scrutiny, or falsely claim that independent verification occurred.
For example, a study may disclose its analytical model, variable definitions, and procedures while limiting access to identifiable participant records.
However, the authors should acknowledge if independent reproduction of numerical results requires access to restricted materials.
The distinction concerns the purpose and consequences of the restriction, not merely whether information is publicly available.
What Is the Difference Between Transparency and Reproducibility?
Transparency concerns the availability of information needed to understand and evaluate research decisions.
Reproducibility commonly concerns obtaining consistent results using the same data, computational procedures, or analytical conditions, although terminology varies across disciplines.
A manuscript may describe its methods transparently without making every underlying dataset publicly available.
Conversely, publishing a dataset without adequate code, documentation, or variable definitions may provide access without sufficient reproducibility.
Researchers should identify precisely what others can verify and avoid treating open access as a substitute for adequate methodological documentation.
Can Restricted Data Still Support Independent Verification?
Potentially. Authorized independent researchers may be able to examine restricted datasets through secure repositories, controlled computing environments, or institutional agreements.
However, the scientific value of such arrangements depends on whether access is feasible, the conditions are reasonable, and the materials are sufficiently complete.
For example, a process that theoretically permits access but rarely approves qualified requests may provide limited practical verification.
Researchers should therefore assess both the formal availability of a mechanism and its actual usefulness.
Restrictions should be proportionate to the risks they address.
What Does Proportionate Openness Look Like?
Proportionate openness means choosing an access arrangement that supports legitimate scientific purposes while limiting unnecessary risks.
Different materials within the same project may warrant different levels of access.
For example, researchers might openly publish analytical scripts, instruments, aggregate findings, and documentation while restricting identifiable records.
However, synthetic datasets and summary outputs may not permit full verification of empirical findings.
The publication should distinguish public reproducibility from verification that requires approved access.
The aim is not to maximize restrictions but to identify the least restrictive defensible arrangement.
What if Open Sharing Is Required by a Journal or Funder?
Researchers should examine the applicable policy and any recognized exceptions for confidentiality, privacy, security, or other legitimate constraints.
Requirements differ across journals, publishers, institutions, and funding programs.
Where restrictions may be necessary, researchers should consult the responsible authority before committing to unrestricted release.
They should not falsely claim that data are publicly available or promise access that cannot be provided.
Where no acceptable arrangement exists, a different publication venue or dissemination strategy may need to be considered.
Can Commercial Confidentiality Justify Restricted Openness?
Commercial or intellectual property obligations may legitimately affect access to certain research materials.
However, proprietary interests should be distinguished from ethical harm prevention.
For example, an organization may restrict a model's source code to protect commercial interests rather than because release would endanger participants.
Researchers should accurately disclose the reason for restriction and explain how it affects independent scrutiny.
Commercial confidentiality should not be used to conceal unsupported scientific claims.
When Might Complete Openness Be Ethically Inappropriate?
Unrestricted disclosure may be inappropriate when it would violate binding confidentiality obligations or materially increase serious foreseeable harm without sufficient scientific justification.
Examples may include identifiable participant records, certain sensitive security details, or information subject to legally required protections.
However, the fact that a topic is sensitive does not automatically establish that every associated method or result must be withheld.
Researchers should identify which specific material creates the risk and consider whether narrower restrictions would suffice.
The considerations discussed in withholding methodological details to prevent harm are relevant when technical information itself is consequential.
Can Excessive Secrecy Also Be Unethical?
Yes. Unnecessary restrictions can obstruct independent scrutiny, limit reproducibility, and make it difficult to identify errors or unsupported claims.
They may also restrict participation in scientific research to institutions with privileged access.
For example, a research team might invoke confidentiality to avoid sharing analytical documentation that contains no identifiable information.
If the restriction lacks a defensible basis, it may undermine accountability rather than protect participants.
Responsible research therefore requires scrutiny of the costs of secrecy as well as the risks of disclosure.
Who Should Decide When Openness Must Be Restricted?
Researchers should identify the authorities responsible for the relevant concern.
Depending on the project, decisions may involve an ethics committee, data protection officer, information security specialist, institutional research governance office, funder, or journal editor.
Where affected communities have legitimate interests in data governance, appropriate consultation may also be relevant.
However, no single authority necessarily resolves every scientific, ethical, legal, and social question.
Researchers should document the decision and distinguish mandatory restrictions from discretionary safeguards.
Should Restrictions Be Reconsidered Over Time?
Where feasible, yes. The reasons supporting a restriction may change.
For example, technical information about a security vulnerability may become less sensitive after the vulnerability has been corrected.
Conversely, emerging reidentification techniques or new combinations of available datasets may change privacy risks.
Researchers should consider whether restrictions require periodic review and what conditions would justify broader access.
However, review schedules should reflect the project's actual obligations and resources rather than impose indefinite monitoring without a clear purpose.
How Should Researchers Explain Restrictions to Readers?
A publication should state which materials are available, which are restricted, and the reasons for those restrictions.
Where controlled access is possible, the relevant conditions should be described accurately.
Researchers should also explain whether restrictions limit independent reproduction or evaluation of particular findings.
However, explanations should not themselves reveal protected information.
Transparency about what cannot be shared is an important part of responsible scientific communication.
What Ethical Principle Should Guide the Decision?
The central principle is to preserve meaningful scientific accountability while preventing unnecessary and reasonably foreseeable harm.
This requires evaluating both the benefits of openness and the costs of restricting access.
Researchers should avoid assuming that every possible risk justifies secrecy or that every scientific benefit requires unrestricted release.
The appropriate arrangement depends on the evidence, applicable obligations, and the effectiveness of available safeguards.
Watch Out
Do not equate open science with publishing everything or responsible research with withholding anything sensitive. Both approaches can become ethically problematic when applied without attention to the actual materials, risks, and opportunities for independent verification.
