03 · What You Need to Know
How to Make Ethical Decisions When Existing Guidance Is Incomplete
Why Can't Research Ethics Guidelines Anticipate Every Situation?
Research ethics guidelines provide standards for protecting participants, evaluating risks, allocating responsibilities, and maintaining the integrity of research. They cannot, however, describe every possible combination of technology, methodology, population, and research setting.
New situations may emerge when researchers use generative AI, analyze digital traces, combine datasets, conduct research across jurisdictions, or introduce technologies that change how personal information can be inferred.
Even familiar research activities can produce unfamiliar ethical questions. An interview study involving adults who voluntarily participate may present different concerns when interviews are conducted through a platform that records conversations for purposes beyond the researcher's control.
Ethical guidance may also differ among institutions, disciplines, countries, and research sponsors. A practice permitted under one framework may require additional safeguards or review under another.
These differences do not necessarily indicate that ethical principles are inconsistent. They may reflect distinct legal responsibilities, institutional mandates, or judgments about acceptable risk.
First, Determine Whether the Guidelines Are Truly Silent
Before concluding that no guidance exists, distinguish a genuinely unaddressed situation from one governed by a broader provision.
A policy may not mention generative AI specifically, yet its requirements concerning confidentiality, informed consent, third-party data processing, or participant welfare may still apply.
Similarly, research involving publicly accessible online information may raise questions addressed by existing privacy principles even when a policy does not explicitly discuss social media or digital ethnography.
Researchers should examine the scope and definitions of the applicable requirements, not merely search for the name of the technology or method.
Relevant authorities may include institutional research ethics policies, national regulations, data-protection laws, professional standards, funding agreements, and conditions imposed by data providers.
When a requirement is ambiguous, an authorized institutional office may be better positioned to interpret its applicability than an individual researcher acting alone.
What Is the Difference Between Legal Compliance and Ethical Acceptability?
Legal permission and ethical justification are related but distinct.
A research activity may be lawful under an applicable data-protection framework yet still raise concerns about fairness, participant expectations, power relationships, or foreseeable harm.
Conversely, a researcher may regard an activity as ethically valuable while facing a legal restriction that prevents it from proceeding in the proposed form.
Legal and regulatory compliance
Whether the research satisfies binding requirements that apply to the activity, institution, participants, or data.
Ethical justification
Whether the researcher's proposed conduct is defensible in light of relevant values, responsibilities, foreseeable consequences, and stakeholder interests.
Ethical reasoning cannot override a binding legal requirement. At the same time, satisfying the minimum legal standard does not necessarily resolve every ethical concern.
Which Ethical Principles Can Help When No Specific Rule Applies?
The Belmont Report identifies respect for persons, beneficence, and justice as foundational principles for research involving human subjects. These principles can help structure ethical reasoning beyond narrowly specified procedures.
Respect for persons directs attention to autonomy, meaningful choice, and appropriate protection for people whose capacity or circumstances may limit independent decision-making.
Beneficence requires careful consideration of potential harms and benefits, including whether risks can be reduced without undermining the legitimate purposes of the research.
Justice raises questions about who bears research burdens, who receives potential benefits, and whether participant selection or research practices unfairly disadvantage particular groups.
Other frameworks, including the Declaration of Helsinki and the CIOMS International Ethical Guidelines, offer additional guidance for research within their respective scopes.
These principles do not function as an automatic decision algorithm. They may point toward competing considerations. Respecting a participant's preference, for example, does not eliminate the need to consider risks to other people.
The researcher's task is to explain how the relevant principles apply to the actual situation rather than invoke them as abstract assurances that the study is ethical.
Identify Everyone Whose Interests May Be Affected
Ethical evaluation should not stop with the people formally enrolled as participants.
Research may affect individuals mentioned in collected records, communities represented in datasets, institutional employees, or people whose information becomes identifiable through data linkage.
For example, students may consent to share their conversations with an AI learning assistant, but those conversations could contain information about classmates or teachers who have not consented.
The researcher should consider whether such information is necessary, whether it can be minimized, and whether its collection or use creates additional obligations.
Power relationships also matter. Students may experience an invitation from their instructor differently from an invitation issued by an independent researcher, even when participation is formally described as voluntary.
Identifying affected interests does not mean every person mentioned in a dataset must necessarily be recruited as a participant. It means that the ethical assessment should recognize foreseeable consequences beyond the immediate research sample.
Evaluate Risks That Existing Guidelines May Not Explicitly Name
Traditional research ethics procedures often address physical, psychological, social, and informational risks. Emerging research methods may introduce unfamiliar forms or combinations of these risks.
AI-mediated research, for example, may create concerns about secondary data use, model memorization, automated inference, platform access, and the disclosure of sensitive information through generated outputs.
The existence or likelihood of these risks should not be assumed without examining the actual system, data flows, contractual arrangements, and technical safeguards.
A practical risk assessment should consider the nature of potential harm, its likelihood, severity, reversibility, and distribution among affected people.
Researchers should also examine whether the same scientific question could be addressed using less intrusive data or a different procedure.
Ethical evaluation is not limited to identifying possible harm. It also involves considering the value of the research and whether proposed safeguards are proportionate to the risks.
What Should You Do When Ethical Principles Point Toward Different Decisions?
Ethical principles may conflict in application. Protecting confidentiality could limit the information available for independent verification. Respecting participant autonomy could permit choices that involve some risk. Preventing possible harm could restrict participation in research that affected communities consider valuable.
When competing considerations arise, identify the specific interests at stake and explain why one proposed arrangement is preferable to reasonable alternatives.
A decision should consider whether the research objective is sufficiently important, whether the risks are proportionate, and whether less burdensome safeguards could address the concern.
There may be more than one defensible response. The relevant question is whether the chosen approach is supported by careful reasoning, appropriate evidence, and the applicable governance process.
Where disagreement remains substantial, making an ethical research decision when reasonable people disagree requires attention to the quality of the justification rather than simply the number of people supporting each position.
When Should You Consult a Research Ethics Committee?
Consultation is particularly important when uncertainty affects participant welfare, consent, privacy, vulnerability, risk classification, or the scope of an existing ethics approval.
A research ethics committee or institutional review board may be able to clarify whether a proposed activity requires formal review, an amendment, additional safeguards, or another institutional authorization.
However, ethics committees differ in jurisdiction and responsibilities. A committee may not have authority to interpret employment law, resolve contractual ownership disputes, or provide definitive technical security assessments.
Depending on the issue, consultation may also be needed with a data-protection officer, information-security specialist, legal office, community representative, or qualified methodological expert.
For research already approved, investigators should not assume that an unfamiliar procedure can be introduced without review merely because the original protocol addressed a related activity.
The need for independent ethical advice even when formal review is not required may be especially strong when a decision has substantial consequences or the researcher has a conflict of interest.
How Should Researchers Handle Conflicting Institutional or International Standards?
Cross-institutional and international research may be subject to overlapping requirements. A university's ethics approval does not necessarily replace another jurisdiction's legal obligations or a collaborating institution's mandatory procedures.
Researchers should identify which requirements are binding, which are advisory, and whether the relevant authorities recognize one another's review processes.
Simply choosing the strictest-sounding rule may not resolve the problem. A requirement intended to protect one interest could conflict with another obligation or create unnecessary exclusion.
The question of how to respond when research ethics standards conflict therefore requires interpretation of their scope and purpose.
When obligations appear incompatible, researchers should obtain guidance from the relevant authorities before proceeding. An individual investigator should not privately decide to disregard a binding requirement.
How Do You Document a Decision When the Guidelines Are Unclear?
Documentation is particularly valuable when the applicable guidance does not provide a straightforward answer.
A useful record should identify the ethical question, relevant policies and principles, affected interests, available options, anticipated risks, consultations undertaken, and the justification for the selected approach.
It should also record unresolved uncertainty, required approvals, safeguards, and any conditions that would trigger reconsideration.
For example, a researcher might decide to use de-identified AI tutoring transcripts only after verifying the platform's data-processing arrangements and obtaining the required institutional authorization.
The record should distinguish verified facts from assumptions. It should not claim that a platform guarantees anonymity merely because direct identifiers were removed.
The process of documenting a difficult ethical research decision can help future reviewers understand the reasoning without treating documentation itself as proof of ethical acceptability.
Can an Ethical Decision Need Reconsideration After Approval?
Yes. Ethical acceptability can depend on circumstances that change during a study.
A platform may introduce new data-retention practices, a previously low-risk dataset may become linkable to identifiable information, or participants may experience harms that were not reasonably anticipated.
Formal ethics approval provides an important governance decision within its scope, but it does not eliminate the researcher's continuing responsibility to recognize new concerns.
Where significant changes occur, investigators may need to pause affected activities, implement immediate protections, or seek an amendment according to institutional requirements.
Recognizing when an approved study needs ethical reconsideration is part of responsible ongoing research oversight.
