Gains:
- Ability to scan for privacy and select the appropriate method (anonymization, closed tool) before uploading sensitive personal data to cloud-based tools
- Ability to clarify the copyright status and cultural sensitivity of a material before use
- Ability to recognize that AI reproduces bias in sources, add a critical framework, and transparently declare the use of AI
History and archive material is not a neutral mass of data. A civil registry may contain the personal information of living people, a court file may contain sensitive private lives, cultural materials considered sacred by a community, works still under copyright. As AI makes it easier to manipulate this material, the consequences of misusing it grow. This unit deepens the ethical framework that surrounds the entire module: privacy, copyright, cultural sensitivity, and transparency and authenticity in the use of AI.
These issues are not technical, but they are at least as important as technical issues; Because a privacy violation, a copyright violation, or ignoring the sensitivity of a community creates both legal consequences and an irreparable loss of trust.
Privacy: the rights of the living and the dead
Archival documents frequently contain personal data: names, addresses, health information, court records, religious/ethnic origin. Some of them belong to living people and are subject to personal data protection legislation (such as KVKK in Türkiye, GDPR in Europe).
- Living persons: Uploading documents containing identifiable sensitive data to public, cloud-based AI tools could be a serious breach. These tools can store data, use it in education, or disclose it.
- Deceased persons: Ethical responsibility remains even as legal protection diminishes; Unnecessarily revealing a person's sensitive past can harm their dignity and that of their family.
- Access restrictions: Many archives impose access restrictions for sensitive documents for a certain period of time. The "efficiency" of AI cannot be a justification for overcoming these constraints.
Caution: Before uploading a document to a cloud-based AI tool, always ask: "Who would be harmed if this content leaked to the Internet?" If the answer is “one of them,” first look for corporate approval, data anonymization, and a local/closed tool if possible. Indiscriminately uploading personal data is irreversible.
Copyright: not every document is publicly available
Just because a document is physically in the archive does not mean that you can use it freely. Copyright is the right belonging to the creator of the work that provides protection for a certain period of time.
- Public domain: Works whose copyright has expired can be used freely. Duration varies depending on country and work.
- Works under copyright: Letters, photographs, unpublished articles may be under copyright; Giving them to AI or publishing them may constitute a violation of rights.
- AI and copyright: Having AI process a copyrighted text means transferring it to the system of a third party (tool provider); This may cause problems both in terms of copyright and contract.
Cultural sensitivity and community rights
Some historical materials touch the memory, belief or identity of certain communities. This includes colonial period records, materials belonging to indigenous/minority communities, and documents considered sacred. The guiding principle here is to respect the voice and rights of the community that produced the material or that it represents. AI's "most efficient" proposal cannot ignore a community's say over its own history. Additionally, AI may unknowingly reproduce biased, derogatory, or exclusionary language in historical sources; It is necessary to be alert to this language.
Transparency and authenticity
In the humanities, originality (that a work is truly the product of your own labor) and transparency (clearly communicating your methods) are core values.
- Declare use of AI: If a transcription, metadata, or translation was produced with the help of AI, record it. Subsequent researchers should be able to verify.
- Limit of ownership of labor: It is against academic honesty to present a comment or text produced by AI as your own original idea.
- Follow institutional and publication guidelines: Many institutions and journals have established rules on how to declare the use of AI.
Subject
basic question
Consequences if it goes wrong
Privacy
Who does this data identify?
Legal violation, harm to privacy
copyright
Do I have the right to use this?
Rights violation, lawsuit
Cultural sensitivity
Whose memory is this?
Loss of trust and respect
transparency
Did I mention that I use AI?
Academic integrity violation
prejudice
Whose voice is distorted/missing?
distortion of history
three mini cases
Case 1 — Uploaded health record. A researcher had a 20th-century hospital record (including information identifying descendants of living people) loaded and transcribed directly into a publicly available AI tool. The institution's data protection officer noticed the situation; the document contained sensitive personal data and the upload was a violation. Lesson: sensitive personal data is not uploaded without a closed/approved tool.
Case 2 — Copyrighted photo. For a publication, a copyrighted photo from an archive was enhanced with AI and published. The photographer's heirs reported a violation of rights. Being in the archive did not grant usage rights. Lesson: the copyright status of each material is clarified before use.
Case 3 — Reproduction of prejudice. A student had YZ summarize a colonial-era record. The summary reproduced without question the derogatory language and one-sided view of the source; The perspective of that community was completely absent. With the warning of his advisor, the student re-examined the source with a critical framework and added the missing audio. Lesson: AI copies the source's bias; It is the researcher's job to add the critical framework and missing voice.
Four copyable templates
1) Pre-installation privacy check:
Perform a privacy check BEFORE loading the following document text into an AI tool: (1) Is there sensitive data (health, religion, ethnicity, address) that identifies living or recent individuals? (2) If so, what sections? (3) Is anonymization possible? The final decision is mine; You mark risky areas. Text: [here]
2) Copyright status evaluation:
Information about the type, estimated date and creator of the material below is as follows: [information]. List what copyright questions I should ask and what I should clarify before use. GIVING final legal judgment; Provide points to check.
3) Bias and missing volume control:
In the historical source summary below: (1) where are the source's own biases/derogatory language reproduced, (2) which group/perspective's voice is missing? Highlight these and suggest how they can be critically framed. Summary: [here]
4) Draft AI use statement:
In one study, I used AI in: [transcription/translation/metadata/summary]. Write a draft "AI use statement" that can be used for academic transparency: in what work, with what tool, how human verification is done. Exaggeration; Be honest and clear.
Weak prompt / Strong prompt
Weak:
Process this archive document and extract its contents for me.
The confidentiality, copyright and sensitivity status of the document is given directly to the AI without being questioned; The risk of breach remains invisible.
Strong:
Before processing this document: mark any areas that may pose a risk in terms of sensitive personal data, copyright and cultural sensitivity. If there is a risky part, I will review my processing method (anonymization, confirmation, closed tool). Extract only approved content. Document: [here]
The difference: risk screening before processing, flagging risky content and reviewing the method prevents ethical violations in the first place.
Common mistakes
- Uploading sensitive data indiscriminately. Personal data will not be uploaded without a closed/approved tool and anonymization if necessary.
- Assuming "it's in the archives, I can use it". Physical access does not confer copyright; the situation is clarified.
- Ignoring cultural sensitivity. The right of a community to have a say in its own history is respected.
- Copying prejudice without question. The one-sided language of the source is addressed with a critical framework, and the missing voice is added.
- Hiding the use of AI. Transparency is a requirement of academic honesty; usage is declared.
In summary
Ethics is the framework surrounding the technical units of this module. Privacy includes protecting sensitive personal data; Clarifying copyright, usage rights; cultural sensitivity, respecting the right of communities to have a say in their own history; Transparency and authenticity require honestly declaring the use of AI. AI also reproduces biases in sources; It is the responsibility of the researcher to add the critical framework and missing voice to this. Efficiency can never be a justification for exceeding these principles. As technical power increases, ethical attention must also increase.
Application task
Select a document you intend to process. First, scan for sensitive data with the “pre-install privacy check” template and flag risky sections. Then remove the questions that need to be clarified before use with the "copyright status assessment". Finally, review a source summary with a “bias and missing volume check.” Make a reasoned decision about whether it is appropriate to upload the document to a cloud-based tool.
checklist
- [ ] I scanned for sensitive personal data before uploading.
- [ ] I have clarified the copyright status of the material.
- [ ] I observed cultural sensitivity and community rights.
- [ ] I addressed the source's bias with a critical framework.
- [ ] I have transparently declared my use of AI.