Gains:
- Understanding the types of plagiarism, self-plagiarism and the ethical status of artificial intelligence produced text
- Ability to write a transparent artificial intelligence use statement in accordance with the institution and journal artificial intelligence use and declaration policies
- Ability to understand the principles of authorship (artificial intelligence cannot be an author), the consequences of data fabrication and ethical violations, and to act responsibly
The entire value of academia is based on trust: the reader of a paper trusts that the author is honest, the data is authentic, the contribution is original, and the sources are cited accurately. When this trust is broken, not only a paper but a career and even the reputation of a field is damaged. Maintaining this trust in the age of AI brings new responsibilities: the limits of plagiarism, the ethical status of AI-generated text, and transparent disclosure. The core principle of this unit is one: the use of AI is legitimate but cannot be concealed; Honesty, transparency and ultimate human responsibility are maintained under all circumstances.
Types of plagiarism and self-plagiarism
Plagiarism is presenting someone else's idea, sentence or data as your own without citing the source. It has several forms. Direct plagiarism: copying the text verbatim. Mosaic/paraphrasing plagiarism: changing words and using the idea without attribution — it is the most insidious type, because it is mistaken for "my own sentence". Self-plagiarism: re-presentation of the author's own previous publication without proper attribution; It is a violation to republish or present the same data as new, even if it is your own text. Salami slicing: artificially dividing a single work into multiple articles.
AI is a two-pronged issue here. On the one hand, having the AI “paraphrase” text would be automating mosaic plagiarism if the original source is not cited — the idea of the AI rewriting is not yours. AI, on the other hand, can help you flag unintentionally uncited passages or risks of self-plagiarism in your own text. The rule does not change: the obligation to cite sources is independent of who or what wrote the text.
Caution: "AI rewrote, now it's my sentence" is not a defense. If the idea is someone else's, attribution is required, no matter how fluently the AI rewrites it. Otherwise, plagiarism has changed the medium but continues to exist.
AI generated text and writing
Common publication ethics guidelines (e.g., the ICMJE — committee of medical journal editors and COPE — committee on publication ethics lines) establish two clear rules. First: AI cannot be a writer. Authorship requires responsibility—defending the integrity of the work, declaring conflicts of interest, being accountable—and a language model cannot do these things. Second: The use of AI is declared. If you used AI in text generation, language editing, or analysis, you transparently disclose this in the method or separate declaration section. The purpose of the statement is not to accuse you; It is to honestly show the reader and editor how the text was produced.
The most serious violations are data fabrication and data falsification. Making the AI say "produce a reasonable finding", "complete the missing data", "show the result stronger" falls directly into this category and is the most serious crime that can end an academic career. This trap is insidious because AI can generate fluent numbers; The red line is absolute: data comes only from actual observation, never produced.
Corporate policy and transparency practice
Each institution and journal's AI policy is different and changing rapidly; some allow broad use while others prohibit certain stages. Your responsibility is to read the current policy of the target journal and your institution (university, thesis regulations) before starting work. The practice of transparency is concrete: record which tool you used, at what stage, for what, and declare it when necessary. This record both protects you and proves your honesty in an audit.
One point should also be clarified: It is wrong to blindly trust AI detection tools. These tools produce both false positives (mistaking human writing for AI) and false negatives; They may unfairly mark natural text, especially by non-native English writers, as "AI-generated." Therefore, it is neither correct to blame a student's text based solely on a detector score, nor to rely on such a tool to say "clean" and skip transparency. The assurance of integrity is not a software score, but your transparent declaration and record-keeping discipline. Ethics is not a measured value, but a practiced responsibility.
three mini cases
Case 1 — Mosaic plagiarism. A student used a paragraph of an article without attribution by having the AI say "paraphrase". The plagiarism software showed low similarity, but the consultant recognized the source of the idea. This was a complete mosaic plagiarism; AI changed the words, but did not eliminate the attribution obligation. Student added the source and learned the lesson: rewriting does not replace attribution.
Case 2 — No to data fabrication. A researcher had missing data in a cell and was tempted to tell the AI to “suggest a reasonable value.” He stopped and thought: this data was fake. Instead, he reported the missingness honestly in the method and stated that he applied an appropriate missing-data technique (imputation). The text remained transparent and no numbers were made up.
Case 3 — Transparent statement. A non-native writer polished the language of his article with AI and also used AI for his introduction draft. He read the journal's policy and added a statement to the methods section, "A generative AI tool was used for language editing and introduction drafting, and all content was verified and approved by the authors." The editor welcomed this; Transparency created trust.
Copiable templates
1) Plagiarism/attribution risk screening:
Mark the sentences of ideas, claims or data in the text below that may require citation but appear to be uncited. Also point out places similar to my previous work that pose a risk of self-plagiarism. It's up to me; you list only the risky places. Text: [paste]
2) AI usage statement:
Write a transparent draft of an AI use statement for an article. Usage:[which tool, which stage: e.g. language editing, draft abstract]. It should be noted that BeyanYZ is NOT the author and all content has been verified by the authors. 2-3 sentences, formal tone.
3) Paraphrasing + attribution (correct usage):
Rewrite the following sentence more fluently BUT remember that this is another source's idea: put a placeholder at the end of the rewrite [REFERENCE HERE] so I can add the source. Don't change the idea. Sentence + source: [paste]
4) Policy checklist:
Create an AI ethics checklist before submitting to a journal: has the institution policy been read, has the journal AI policy been read, has usage been recorded, has a statement been written, has no data been fabricated, have all attributions been verified. Make it short and bookmarkable.
Weak prompt / Strong prompt
Weak prompt:
Rewrite this paragraph so that there is no plagiarism. [paragraph]
This does not solve the attribution problem; If the idea belongs to someone else, rewriting is still plagiarism.
Powerful prompt:
The following paragraph contains the idea of another study. Help me summarize it in my own words BUT (1) leave room to cite the source, (2) don't distort the idea, (3) show me space where I can add my original interpretation. Source: [imprint]. Paragraph: [paste]
Status
Ethical status
correct behavior
Language correction with AI
legitimate
Declare
Writing an idea without attribution with AI
Plagiarism
Add citation
Making AI produce data
serious violation
never do
Using own old text without attribution
self-plagiarism
show source
Show AI as author
unacceptable
Human writer + declaration
Common mistakes
- Substituting rewriting for attribution. If the idea is someone else's, attribution is required even if AI rewrites it.
- Fitting/strengthening data. The most serious violation; It could end a career.
- Hiding usage. When it occurs, it is considered a breach of integrity.
- Ignoring self-plagiarism. Attribution is required, even if it is your own text.
- Not reading the policy. Institutional and journal rules are different and binding.
Tip: A simple self-check question: “Is the idea/data in this sentence really mine, or does it come from another source?” If the answer is "else", reference; If "produced", stop. These two questions prevent most plagiarism and fabrication while you're writing.
In summary
Academic confidence; is maintained with honesty, authenticity and transparency. Plagiarism does not relieve the obligation of attribution if the idea belongs to someone else, even if the intermediary is AI; Self-plagiarism and salami slicing are also violations. AI cannot be the author, data can never be fabricated, and usage is transparently declared. Read institutional and journal policies in advance and record your use. The shortest rule: idea depends on the source, data depends on the truth, use depends on the statement, responsibility depends on the person.
Application task
Take your own example of a text. Scan for citation/self-plagiarism risk with AI and evaluate the marked areas yourself; If you find at least one uncited idea, add a source. Then write a transparent AI usage statement suitable for your own use. Finally, find the AI policy of your target journal and institution and note its three articles.
checklist
- [ ] Have I cited anyone else's ideas/data?
- [ ] Have I checked my own previous work for the risk of self-plagiarism?
- [ ] Am I sure I'm not fabricating/inflating any data?
- [ ] Have I disclosed my use of AI in a transparent statement?
- [ ] Have I read the institution and journal AI policy?