Academic Citation
Exercise on citation practice in historical research
This exercise was translated automatically from the German original and can contain errors. Consult the original version in case of doubt.
Overview and Didactic Goal
This exercise teaches the fundamentals of academic citation in historical research, with a particular focus on the critical use of generative AI. You learn how citations work as a methodological tool that makes historical arguments traceable and verifiable. You also learn how you shape AI-assisted work on text so that the chain of evidence stays intact.
The exercise has two parts. The first part covers the citation of sources and literature. The second part covers the disclosure: the declaration in which you state your own AI use. This exercise closes both learning paths.
At a Glance
| Item | Value |
|---|---|
| Duration | about 90 minutes |
| Level | Beginner |
| Core path | exercise 1 → exercise 2 → exercise 3 |
| Optional deepening | the variant with an article behind a paywall |
| Hand-in (minimum) | One disclosure of the AI use. AI log: 4 entries. |
All three exercises belong to the core path. The optional deepening costs about 20 minutes more.
Exercise 3 works with the AI log from the earlier exercises. Without this log you cannot write a disclosure.
Do you have no log yet? Work through the exercise Prompt Engineering first. You set up your log there.
What You Need
Subject knowledge
- A basic understanding of historical research methods
- Basic knowledge of work with generative AI, above all prompting
If prompting is new to you, we recommend that you work through the exercise Prompt Engineering first.
Capabilities of the AI system
The system must be able to do the two following things:
- Search the web and retrieve an article through its DOI. You need this in exercise 1 and in exercise 2.
- Read a file that you upload. You need this when the system does not retrieve the article itself.
Other tools and access
- A reference manager. You use it to compare the data of the AI with a checked import.
- The test article as a PDF file. The article is freely accessible, and you need no institutional access for it.
- Your AI log from the earlier exercises.
- The binding rule of your institution on AI use. You find this rule with the procedure in this exercise.
You can do every exercise with AI systems from different makers. What a system can do depends on the product and on your account. Settle the five following points before you start.
Cost
A free account is enough for short text tasks. For some capabilities you need a paid account. This mostly concerns the search on the web, the reading of long files and the built-in reasoning mode. Check first whether your university provides access.
Tools
Check which tools your account unlocks. Can the system search the web? Can it read a file that you upload? Can it call external services? Each exercise names the capabilities that it requires.
The section Check and classify capabilities explains five newer capabilities. It also describes what each capability means for your work with sources.
Memory
Some systems carry over statements from earlier sessions or from a project context. Check the settings for memory, projects and chat history. A new session therefore guarantees no empty context.
Account: institutional or private
Check which account you work with. The rules differ in three points.
| Point | Account of your institution | Private account |
|---|---|---|
| Contract | Your university concludes it and shares the liability. | You conclude it yourself and carry the liability alone. |
| Training | A contract often excludes the use of your inputs. | Many providers use inputs for training. |
| Permitted material | The university rules which data you can enter. | You have no such rule and carry the risk yourself. |
Ask the IT office or the library of your university when you cannot find the rules. Record the answer in your AI log.
Do not upload material that you have no right to pass on. An upload leaves your computer and you cannot take it back.
These materials stay outside the AI system:
- Personal data of living persons, for example from interviews, from private papers or from holdings under a protection period.
- Digitized items whose terms of use forbid a disclosure.
- Unpublished texts of other persons, for example manuscripts or reviews.
- Licensed full texts, when the license of your library forbids a disclosure.
In case of doubt, work with a short excerpt or with an anonymized version. You can also use a local model, because the material then stays on your computer.
Law: data protection and disclosure
Two areas of law concern your work directly. This section gives an orientation and no legal advice.
Data protection. In the EU the General Data Protection Regulation applies. In Switzerland the revised Data Protection Act applies (Verordnung (EU) 2016/679 Zum Schutz Natürlicher Personen Bei Der Verarbeitung Personenbezogener Daten (Datenschutz-Grundverordnung) 2016, Bundesgesetz Über Den Datenschutz (DSG) 2020). Both laws protect personal data of living persons. The text of the law itself is binding. Article 6 GDPR requires a legal basis for every processing operation, and an upload is a processing operation.
The role of the provider decides the classification. A provider can act as a processor and process your data only on your behalf. Article 4 point 10 GDPR expressly excludes a processor from the third parties. This role needs a contract under Article 28 GDPR. Without such a contract, a provider uses your inputs for its own purposes. An upload is therefore not always a disclosure to a third party, but without a contract you must assume that it is. Check the contract that applies to your account.
Research is subject to safeguards, but it gets no exemption. Article 89(1) GDPR requires appropriate safeguards, for example data minimisation and pseudonymisation. Article 89(2) GDPR permits the legislator to derogate from single rights of the data subject. The information of the data subjects can also fall away, namely under Article 14(5)(b) GDPR. That applies only when the information proves impossible or would involve a disproportionate effort. Research therefore does not release you from the information duty in general. Settle your individual case with the data protection office of your university.
The European Data Protection Board presented a guideline on research in 2026 (European Data Protection Board 2026). This guideline was in public consultation until June 2026. It is therefore not applicable law, but an aid to interpretation. For deceased persons, the protection periods and the rules of use of the archive apply.
Disclosure. The AI Act of the EU governs the transparency of generated content (Verordnung (EU) 2024/1689 Zur Festlegung Harmonisierter Vorschriften Für Künstliche Intelligenz (Verordnung Über Künstliche Intelligenz) 2024). Article 50 applies from 2 August 2026. It separates two addressees: the provider of an AI system and the deployer. You are a deployer as soon as you use an AI system professionally. For purely private use you are not a deployer, under Article 3 point 4.
As a deployer you carry two separate duties from Article 50(4).
- Deep fake. If an AI system generates or manipulates an image, an audio recording or a video, and the content is a deep fake, then you disclose that. Under Article 3 point 60, a deep fake resembles existing persons, places or events and falsely appears authentic. A generated image without this resemblance to reality does not fall under this duty.
- Text on public matters. If you publish generated text to inform the public on a matter of public interest, then you disclose that. The duty falls away when a person reviewed the text and holds the editorial responsibility for the publication.
Article 50(4) therefore requires no disclosure for every generated image and for every generated text. Article 50(2) addresses the provider instead. The provider must mark generated content in a machine-readable format. This duty does not lie with you, and it does not replace your own disclosure.
The legal disclosure and the academic disclosure are two separate duties. Your university requires a declaration even when the law requires no disclosure. The exercise Citing shows how you declare your AI use.
Record your answers to these five points. They belong to the description of your work environment.
Learning Objectives
After the exercise you can:
- explain the epistemic and ethical functions of academic citations in historical research
- tell common knowledge (widely known, uncontested background knowledge that needs no reference), primary sources and secondary literature cleanly apart
- decide consistently when a citation is necessary, and when it is not
- apply footnote citation practice to an established standard, for example the Chicago Manual of Style
- use citations deliberately to support, to embed and to qualify historical arguments
- use generative AI as a heuristic tool, without damage to the chain of evidence
- state your own AI use in a disclosure, and derive this disclosure from your AI log
- find the binding rules of your institution and of your target journal yourself
What Are Academic Citations?
Academic citations are the formalized system that references information, ideas, data and interpretations. In historical research they work as a methodological control mechanism: they make arguments verifiable, reconstructable and open to critique, because they give readers access to the base of evidence.
Citations tie arguments to:
- primary sources (contemporary documents, files, letters, minutes and so on)
- secondary literature (later scholarly interpretations)
Without this link, historical texts lose their scholarly character and become mere narrative. (The Turing Way Community 2022)
Purposes of Citation
1) Academic Integrity and Attribution
- It prevents plagiarism, because intellectual debts become visible.
- It states the authorship and the prior work.
- It stabilizes credibility through responsible practice.
2) Verifiability and Transparency
- Readers can find and evaluate the sources.
- Citations record the chain of evidence (source → analysis → claim).
- The work effort also becomes visible (work in archives and with data).
3) Position in the State of Research
- Citations locate statements in the discourse.
- They mark agreement, delimitation, controversy and connection.
4) Argument Based on Evidence
- Primary sources: the empirical foundation.
- Secondary literature: context, theory, method, state of the debate.
5) Discursive and Technical Functions of Footnotes
- They contextualize or translate a quotation.
- They carry comments from source criticism.
- They point to counter-evidence and to alternative readings.
- They give reading recommendations without an overload of the main text.
Rule to remember: citations do not guarantee “truth”, but inspectability.
When Must I Cite?
Every statement that goes beyond general background knowledge needs a citation.
Always cite:
- direct quotations (quotation marks and a reference)
- paraphrases (the idea stays borrowed, even when the wording is new)
- arguments and frames of interpretation that you adopt
- data, statistics and operationalized measurements
- indirect adoptions (a primary source through secondary literature → a double attribution)
Exception: common knowledge. Widely known, uncontested, trivial facts need no reference. But: as soon as you qualify, sharpen, quantify or enter contested ground, a citation is due.
How Do I Cite?
Use a reference manager, for example Zotero, and follow an established standard. In historical research this is mostly the Chicago Manual of Style in its 18th edition of 2024 (The University of Chicago Press Editorial Staff 2024). The complete work costs money. The short Chicago-Style Citation Quick Guide is freely accessible (ENGLISH only). A German-language alternative are the guidelines of infoclio.ch (GERMAN and FRENCH only).
Different conventions can apply, depending on your institution or on the context of publication. Inform yourself in advance and stay consistent.
Declare Your AI Use
A citation and a disclosure are two different things. A citation gives the reference for a single passage. A disclosure, also called a declaration, describes your work process. It says whether and how you used generative AI. A piece of work often needs both.
From the AI Log to the Disclosure
In the other exercises you keep an AI log. The log is your private record of work. It holds prompts, answers, verification steps and decisions. The disclosure is the public statement. You condense the log into a few sentences.
| Feature | AI log | Disclosure |
|---|---|---|
| Addressee | you, and your supervisor on request | all readers |
| Extent | complete, all entries | short, mostly 5 to 15 sentences |
| Place | your own file or an appendix | a fixed place in the work |
| Purpose | reconstruction of the work process | transparency and responsibility |
| Time | continuous during the work | at the end, derived from the log |
Rule to remember: without a log you cannot write a truthful disclosure.
When Must I Disclose?
Check the rule of your institution before you rely on a rule of thumb. Many universities require a declaration for every use of AI, even for a spelling check.
Disclose the AI use as soon as it is substantial. Substantial means that the AI shaped your result in its content. The Living Guidelines of the European Commission count the literature review, the interpretation of analyses, the formulation of research goals and the development of hypotheses among these uses. A pure correction of text does not count there as a substantial use (European Commission, Directorate-General for Research and Innovation 2026).
Disclosure and Legal Marking Are Two Duties
The academic disclosure and the legal marking have different addressees. Keep the two apart.
| Question | Academic disclosure | Legal disclosure |
|---|---|---|
| Basis | the rule of your institution or of your journal | the AI Act of the EU, Article 50(4) (Verordnung (EU) 2024/1689 Zur Festlegung Harmonisierter Vorschriften Für Künstliche Intelligenz (Verordnung Über Künstliche Intelligenz) 2024) |
| Applies | from every substantial use | from 2 August 2026, for a professional use |
| Covers | every text type, a seminar paper included | a deep fake, and generated text published to inform the public on matters of public interest |
| Addressee | your scholarly community | the public |
| Exception | none, when the institution requires a declaration | text that a person reviewed and for which that person holds the editorial responsibility |
The regulation therefore requires no marking for every generated image and for every generated text. A third duty does not concern you: Article 50(2) obliges the provider to mark generated content in a machine-readable format.
The exception in the regulation does not release you academically. You disclose your AI use even when the law requires no marking.
What Belongs in a Disclosure?
The German Research Foundation names four items. Researchers should disclose whether and which generative models they used, for which purpose and to which extent (Deutsche Forschungsgemeinschaft 2023). Add two more items, so that the chain of evidence stays visible:
- Whether: did you use generative AI? Answer with yes or no.
- Which: name the tool, the version and the provider. Name the period as well.
- Purpose: name each task on its own, for example the outline, the literature search or a translation.
- Extent: name the chapters or sections concerned. Give numbers where that is possible.
- Verification: describe how you verified the results.
- Responsibility: declare that you answer for the content.
Name also what you did not use AI for. This item protects you from a misunderstanding.
Where Does the Disclosure Stand?
The place depends on the type of text. The following table names the usual places. The rule of your institution or of your journal always takes precedence.
| Type of text | Place of the disclosure |
|---|---|
| Seminar paper, proseminar paper | a signed declaration at the start or at the end, often on a form of the institution; the details in a footnote or in an appendix |
| Thesis (bachelor, master, dissertation) | the form of the faculty under the examination rules, plus a short section in the method part |
| Article, book chapter, monograph | the method part or the acknowledgements, following the author guidelines of the publisher |
| Talk, poster, blog post | a short note at the end or on the last slide |
The Institute of History of the University of Bern, for example, requires a signed AI declaration with every seminar paper. A missing declaration counts there as deception, like a plagiarism (Historisches Institut, Universität Bern 2026). Journals mostly ask for the item in the method part at submission (International Committee of Medical Journal Editors 2026).
Find the Binding Rules
This exercise explains a procedure. Only the rules of your institution and of your journal are binding. Search for these rules before you hand in a piece of work.
- Open the website of your institute. Search for “plagiarism”, “declaration of authorship” or “AI”.
- Download the template for the declaration of authorship. Check whether it holds an AI declaration.
- Read the study guide of your subject. Many guides hold a section on generative AI since 2023 (Luther et al. 2026; Leitfaden Geschichtswissenschaft 2025).
- Check the examination rules when you write a thesis.
- For a publication, open the author guidelines of the journal. Search there for “AI” or “generative”.
- Note the place where you found the rule, and the date, in your AI log.
- Ask your supervisor when a rule stays unclear. Ask before you hand in.
Cite AI Output
The disclosure describes the process. You need a citation in addition, as soon as you take AI-generated text into your work, verbatim or in substance. The Chicago Manual of Style covers this case in section 14.112 since its 18th edition (The University of Chicago Press Editorial Staff 2024).
Three rules apply there:
- Give the reference for the AI content at the place, in the text or in a footnote.
- Do not put the AI content in the bibliography. One exception applies when a public link to the output exists.
- Name the provider, the prompt, the tool with its version, and the date.
Model for a footnote:
1. [Provider], answer to "[your prompt, shortened]", [tool and version],
[date of generation], [public link, if one exists].An AI text is neither a source nor secondary literature. It supports nothing. Treat every statement of the AI as a hypothesis for a search that you must verify externally.
Model Example: A Disclosure in a Seminar Paper
The following example stands at the end of the work, before the bibliography. It supplements the form of the institution, and it does not replace it. Replace every item in square brackets with your own.
## Disclosure of the AI use
I used generative AI for this work. I used [tool, version, provider] for it
between 3 and 14 March 2026.
Purpose and extent:
- Outline: I had three variants of an outline generated. I adopted variant 2
for chapters 2 to 4, after a heavy revision.
- Literature search: I asked for secondary literature on Swiss European policy.
Of 12 suggested titles, 5 did not exist. I checked the other 7 titles in the
library catalogue. I cite only titles that I checked.
- Language: I had chapters 1 and 5 checked for grammar and spelling.
I used no AI for the interpretation of the sources and for the translation of
the French documents.
Verification: I checked every statement of content from the AI against the
source or against the secondary literature. I compared all verbatim quotations
with the original (wording and page number).
Responsibility: the content of this work is my own achievement. I answer for
all statements and for all references.
The complete AI log holds 14 entries. It exists and can be inspected on request.
[Place], [date]
[Name]For an article, a short paragraph in the method part is enough:
For the preparatory work on this article I used a generative language model
([tool, version, provider], accessed in March 2026): to structure the argument,
to summarize 40 newspaper articles and to revise the language. I checked all
summaries against the original articles. The model analysed no sources and
contributed no interpretation. The authorship and the responsibility for this
text lie with me alone. Prompts and answers are deposited at [link].Avoid the sentence “This work was produced with the support of AI.” It names neither the tool nor the purpose nor the extent nor the verification. Such an item is not verifiable and does not meet the requirements.
Structure of the Exercise
- Exercise 1: test and compare automated citation.
- Exercise 2: check AI summaries for correct citation and fidelity to the source.
- Exercise 3: write your own disclosure from your AI log.
Exercises 1 and 2 work with an open access article. You need no institutional access for it:
Andreas Fickers, “Update für die Hermeneutik. Geschichtswissenschaft auf dem Weg zur digitalen Forensik?”, Zeithistorische Forschungen 17 (2020), issue 1, pp. 157–168, https://doi.org/10.14765/zzf.dok-1765 (Fickers 2020) (GERMAN only).
The full text and a PDF are freely available at https://zeithistorische-forschungen.de/1-2020/5823. Download the PDF before you start. You need it to check the AI answers against the original.
Variant with institutional access: repeat both exercises with an article behind a paywall, for example https://doi.org/10.1093/pastj/gtae018. Compare how the answers of the AI differ.
Exercise 1: Automated Citation
Cite the article <https://doi.org/10.14765/zzf.dok-1765> according to the Chicago Manual of Style (18th edition).
Build in addition an entry in a format that I can import into Zotero directly.A system with a web search normally retrieves the article itself through the DOI. Check whether it really did so.
Only in the exceptional case, when the AI gets no access to the article, upload the article as a PDF file. You can also copy the content and the metadata to the start of the prompt and separate them visually, for example with “““. Example:
"""
DOCUMENT CONTENT
"""
"""
DOCUMENT METADATA
"""
YOUR PROMPTCompare the answer with your manual citation, with the official citation recommendations on the publisher page, and with the DOI import in Zotero. Test further sources of secondary literature (other articles, monographs).
- Do all items agree?
- Do you find systematic deviations, for example in page numbers, editors or journal titles?
Exercise 2: AI-Assisted Summary with Original Quotations
Summarize the article <https://doi.org/10.14765/zzf.dok-1765>.
Use only original quotations from the article, and cite them according to the Chicago Manual of Style (18th edition).Check the answer against the original article. The PDF is freely available, so you can look up every quotation directly.
- Do the quotations agree with the original text (wording, page number)?
- Did the answer render the context of the quotations correctly?
Exercise 3: Declare Your Own AI Use
Goal
You write a complete disclosure for a piece of your own work. You derive this disclosure from your AI log.
Task (Manual)
- Open your AI log from the earlier exercises.
- Search for the binding rule of your institution with the procedure in the section Find the binding rules.
- Assign each entry of the log to a purpose, for example search, outline or language.
- Write the disclosure after the model example. Use all six items.
- Name at least one task for which you used no AI.
Task with AI
Here is my AI log:
"""
[INSERT LOG]
"""
Here are the rules of my institution:
"""
[INSERT RULES]
"""
Formulate a draft of a disclosure of the AI use from this material.
Name whether, which tools, for which purpose, to which extent, how verified
and who is responsible. Invent nothing. Mark every item that the log lacks
with [MISSING].Work Assignment (Reflection)
Compare the draft of the AI with your own version.
- Which items did the AI invent or embellish?
- Which gaps in the log did the AI make visible?
- Does the draft meet the rule of your institution?
Correct every wrong item by hand. You answer for the disclosure, and not the AI.
Learning Outcome
You take four results away from this exercise:
- a comparison of AI citations with the checked citation, from exercise 1 and exercise 2
- a finished disclosure for a piece of your own work, from exercise 3
- the place where you found the binding rule of your institution, noted in the AI log
- an AI log with at least 4 entries from this exercise
Further Resources
- DFG: Statement on the influence of generative models – Short foundational text on disclosure and authorship (Deutsche Forschungsgemeinschaft 2023) (GERMAN only).
- European Commission: Living Guidelines on the Responsible Use of Generative AI in Research – Recommendations for researchers, third version of May 2026.
- ICMJE Recommendations – Rules of many journals on the disclosure at submission.
- Historisches Institut Bern: Plagiate und KI-Gebrauch – Example of an institutional rule with a template for an AI declaration (GERMAN only).
- Stanford HAI: AI-Detectors Biased Against Non-Native English Writers – Report on a study that tests AI detectors. The detectors often misclassify texts of students who are not native speakers of English as machine-written.
- The Turing Way – Handbook for reproducible, ethical and collaborative research.
Bibliography
Citation
@inreference{mähr2025,
author = {Mähr, Moritz},
title = {Academic {Citation}},
booktitle = {Critical AI Literacy for Historians},
date = {2025-12-29},
url = {https://maehr.github.io/critical-ai-literacy-for-historians/en/exercises/citing.html},
langid = {en}
}