Key Takeaways:
- By 2026, 90% of online content could be AI-generated, raising concerns about quality and data pollution.
- Copyleaks found nearly 60% of ChatGPT AI-generated outputs contain plagiarism.
- Varying plagiarism rates vary across subjects, with Physics having the highest rate, and Theater, Humanities, and English having the lowest.
- Detecting AI-generated content can help students adhere to academic honesty agreements.
AI has begun flooding the Internet with tools readily accessible on platforms like Google. If you think it feels like AI is everywhere, that could be because, in many ways, it is. The unprecedented amount of AI-generated content is saturating the Internet. Some educators have begun adding AI-generated content to their classroom policies alongside plagiarism agreements to help reduce violations of academic integrity.
But, is using ChatGPT plagiarism?
While AI-generated content isn’t considered direct plagiarism, we dug deeper to see if Chat GPT plagiarizes, and the results might surprise you. Before we get into that, let’s talk about how plagiarism is defined.
How We Define Plagiarism
Plagiarism is when a writer (author, student, employee, etc.) uses the work of another artist without citations or proper credit.
When some people think of plagiarism, they may think of copying and pasting the work of another author and submitting it as their own. While this is one form of plagiarism (identical text), there are other forms of plagiarism that can carry the same consequences.
These forms include minor changes and paraphrased text.
Now that you have a sense of what plagiarism is, we’re going to dig into how ChatGPT generates content.
How Does ChatGPT Generate Content
ChatGPT and other AI models generate content through machine learning. In short, AI programs are fed large amounts of text or datasets (and when we say “large amounts,” we mean massive amounts). The programs then analyze the datasets, and “learn” by identifying pattern recognition,syntax, word choice, and language, according to Conductor.
ChatGPT doesn’t just copy and paste from multiple articles; instead, it analyzes the text and essentially “predicts” the order in which the words should go to create a piece of content.
Because AI models like ChatGPT are LLMs, they can’t write without data being input into the system.
But if it doesn’t know what it’s doing, can it really plagiarize?
Evidence of Plagiarism in ChatGPT Outputs
Here at Copyleaks, we’ve always been interested in how plagiarism works, and as AI became more and more popular, we wanted to see if it was actually delivering unique content and determine the degree to which AI-generated content is original and free of potential plagiarism.
Our process was simple: We asked GPT3.5 to write 1,045 pieces of content, averaging 412 words, across all outputs, in 26 subjects that included Accounting, World History, Art, Physics, Law, Mathematics, Music, Philosophy, and Social Science.
Here’s what we found:
Types of Plagiarism Found
*Identical Text: A one-for-one copying of someone else’s text that is passed off as your own
**Minor Changes: Content with minor alterations to the source material, such as altering a verb within a sentence (e.g., slow to slowly)
***Paraphrased Text: Putting someone else’s idea into your own words without crediting the original source
Copyleaks then conducted an in-depth analysis to gauge the specific outputs with the highest levels of identical text, minor changes, and paraphrasing across all 26 subjects.
Minor Changes: Content with minor alterations to the source material.
The individual GPT3.5 outputs with the highest percentages of minor changes were in Physics and Psychology, where our plagiarism checker found that 25.2% of each respective output contained minor changes without proper credit to source material.
Paraphrased Text: Putting someone else’s idea into your own words without crediting the source material.
The individual GPT-3.5 output with the highest percentage of paraphrasing was in Computer Science, where a surprising 80.7% of the text was paraphrased. This was followed by an individual Physics output where 76.3% of the text was paraphrased.
Identical Text: One-for-one of someone else’s text with no citations, passed off as your own.
The analysis found that the individual GPT3.5 output with the highest percentage of identical text was in Physics, where a whopping 27.0% of the text was identical to other sources. This was followed by an individual Chemistry output where 24.7% of the text was identical to other sources without citing the source material.
Similarity Scores: Copyleaks-specific scoring method aggregating the rate of identical text, minor changes, paraphrased text, and more.
For context, a score of 0% signifies that all the content is original, whereas a score of 100% means none of the content is original.
How did ChatGPT score?
Highest Average
The subject with the highest average Similarity Score was Physics, with 31.3% unoriginal content. Psychology came in second at 27.7%, and Science was third with 26.7% unoriginal content.
Lowest Average
The subjects with the lowest average Similarity Score were Theater at 0.9%, Humanities at 2.8%, and English Language at 5.4%.
Highest Overall
The analysis found that the individual GPT3.5 output with the highest Similarity Score was in Computer Science with an astounding 100%. You read that right. ChatGPT generated a piece of content that was completely unoriginal.
As AI-generated content expands and continues to saturate the Internet, staying informed about what’s in that content, especially before we publish it or submit an assignment.
The data shows nearly 60% of AI-generated content contains some form of plagiarism, and this could lead to potential academic and career consequences. As a student or employee, using AI-generated content can be dangerous, and suspicion of use of AI-generated content can be an issue as well.
Full-spectrum protection that includes AI and plagiarism detection can help you ensure your work is original, and in compliance with copyright and licensing laws.
How to Detect & Prevent Plagiarism in AI
Avoiding plagiarism may seem overwhelming when you’re first introduced to citations. For students, learning various style guides as required by your institution can help you along the way. For employees and employers, creating a brand guide that include the writing style for citations can help your company thrive.
For anyone looking to submit work for publication or the classroom, knowing that you’re delivering original content is key. Using a free online plagiarism checker and AI detector can give you the peace of mind that your work is unique and within compliance of copyright laws.
Request a custom Copyleaks demo and see how the world’s top enterprises ensure trust and transparency.
Responsible Use of ChatGPT
Despite ChatGPT’s potential shortcomings in plagiarism, there are some benefits to using ChatGPT as a tool rather than as a writer.
ChatGPT is great for generating ideas and helping with the brainstorming phase of the writing process. However, as with anything, double-check the work. For example, if you ask ChatGPT a question about the life of squirrels in the wild, no matter what the answer, tell ChatGPT to cite its sources. If you’re given a link, dig deeper to see if the source is credible. After all, you don’t want to submit a paper that says squirrels love calamari unless you get that information from a reliable source.
Always consider that ChatGPT is not a reliable source of information. It is a computer program. Be skeptical of the output, as it may sometimes be incorrect.
Add original thought to your work, and don’t forget to credit your source material, always.
Remember, it’s essential to always comply with the guidelines of your institution, university, or employer. If an instructor is comfortable with you using ChatGPT to generate ideas, go for it. If you are to stay away from ChatGPT at all costs, avoid it!
When it comes to ChatGPT, remember that you are the writer, not another artist, not another writer, and NOT ChatGPT.
Further Resources On Generative AI and Plagiarism
- July 8, 2025
Third-Party Studies Continue to Confirm Copyleaks AI Detector as the Most Accurate
Independent third-party research continues to confirm the Copyleaks AI Detector as one of the most accurate solutions available.
- March 3, 2025
Copyleaks Research Reveals AI Has Unique Stylistic Fingerprints
Copyleaks research uncovers unique AI fingerprints, accurately identifying which AI model generated a text with 99.88% precision.
- September 17, 2024
Why We Need Governance and Compliance in Source Code
Copyleaks CEO discusses why source code governance and compliance are crucial in today’s AI-driven world.
- September 5, 2024
AI Detector Accuracy for Non-Native English Content
Copyleaks study shows the high accuracy of AI Detector for non-native English texts, addressing bias concerns and exploring educational implications.