Does Turnitin Detect if You Use ChatGPT?

openAI chatGPT

Does Turnitin Detect if You Use ChatGPT?

Key Points

  • Turnitin can detect ChatGPT and other AI models with a 98% accuracy rating.
  • Turnitin’s AI detection covers ChatGPT 3.5, ChatGPT 4, and Google’s Bard.
  • Turnitin can screen content to determine if ChatGPT was used before analysis.
  • Other plagiarism detectors that can detect AI include OpenAI Text Classifier, GPTRadar, Copyscape, and GPTZero.

Can Turnitin detect ChatGPT? ChatGPT has made a massive splash on the landscape of writing since its release to the public last year. As you can imagine, this has led to a massive influx of AI-generated works hitting the likes of professional and academic writing.

ChatGPT isn’t truly intended for the likes of student-created essays, research papers, and other school-assigned materials. As such, plagiarism and AI content detectors have been on the rise to counteract this surge of materials. Can Turnitin detect ChatGPT? Read on and find out.

What Is ChatGPT?

ChatGPT is a language learning model, or LLM, that uses countless text inputs to shape its model. The user interface is more akin to a conventional chatbot, albeit far more robust than your typical chatbot. ChatGPT works on the basis of prompts, specializing more in written works than conversation.

ChatGPT is developed and maintained by OpenAI, which specializes in developing AI products. The company itself has caught the attention of the likes of Microsoft, which serves as one of its primary investors and leverages the technology for its own uses, including Bing.

Its Intended Use Case

does turnitin detect chatgpt
ChatGPT can be used to generate written content with a simple prompt


ChatGPT’s intended aim runs against the actual use case of most plagiarism detectors. ChatGPT itself can be used for generating high-quality content for the likes of blog posts and product descriptions.

The ethical impact of this is something rarely discussed, however. ChatGPT uses countless written sources by actual people to form the basis of its language model.

Does Turnitin Detect ChatGPT?

So, before diving into the full details, Turnitin does indeed detect ChatGPT. It can also detect the likes of Google’s Bard. If you’re thinking of using ChatGPT for your next college essay, you might want to rethink that decision.

The AI Detection Model

The AI detection model employed by Turnitin functions in the same way as its usual plagiarism detection. Each sentence of a given work is analyzed and assigned a score. Since ChatGPT and other AI models use pre-written content, it can pick up on these fragments fairly quickly.

Turnitin claims its AI detection model has a 98% accuracy rating, which is quite stellar. There is also the simple fact that AI has a specific voice and tone to its written works.

The AI Models That Turnitin Can Detect

does turnitin detect chatgpt
The recently introduced Google Bard is also detected with Turnitin’s AI detection methods.

Turnitin has support for ChatGPT 3.5, ChatGPT 4, and Google Bard. Other AI models are likely quickly detected since most of these AI chatbots use similar parameters for training. Turnitin’s AI detection was released on April 4th, 2023.

Turnitin has a few different tools with support for AI detection. Depending on the tools deployed by your organization or school, you’ll see support for it in Turnitin Feedback Studio, Turnitin Originality, and Turnitin Similarity. Turnitin’s deployment of AI detection is comprehensive and covers pretty much the entire product line.

How the Detection Works

Like any other written work submitted to Turnitin, each sentence is evaluated for AI generation. Once a piece is submitted, each sentence is given a score between 0 and 1. A score of 0 for a given sentence means it matches human writing patterns. A score of 1 indicates that AI was used to generate it.

These scores aren’t taken in a vacuum but are instead weighted and averaged over the whole of the piece. While ChatGPT can generate vast swaths of text with a simple prompt, it has patterns to its generation. Turnitin’s AI detection methods have analyzed these patterns thoroughly.

Since so much of ChatGPT’s modeling data is taken from internet sources, sentences have a certain method of construction that relies on the next probable word in a sequence.

AI Screening

Turnitin doesn’t just detect AI works — it can also screen them. Before you run one of the AI detection tools that Turnitin offers, it can screen the whole of the content to determine whether ChatGPT was used.

Screening before analysis gives a solid foundation of what to expect for a given written work, especially if you suspect a student or employee of using AI to generate text.

The Ethics of Using AI for Academic and Professional Works

does turnitin detect chatgpt
It might be a difficult process, but you are better off writing in your own words rather than relying on AI.

AI as a replacement for written works has dubious legal and ethical standing. If you’re in a position where you are expected to turn in original written works, then an AI-generated piece is a clear violation of the trust placed on you by your instructor or employer.

That isn’t to say you should avoid AI models like ChatGPT. There is certainly value in using it as a tool, provided it is permitted in your field of work. Given the ethical dilemmas arising for students and professional writers alike, it is best to use your own words that reflect your knowledge instead.

Other Plagiarism Detectors that Can Detect AI

While AI models have overtaken the news cycles, AI detectors are rising to the challenge.

OpenAI Text Classifier

In a curious move, OpenAI has its own AI detector. Once you’ve fed the Text Classifier a minimum of 1,000 characters, it can determine whether an AI or human wrote a given piece. The OpenAI Text Classifier isn’t always accurate, but it is free.

If you’re running a blog or tutoring, you might not have the desire to carry a monthly sub to check the content. However, this tool is free. The accuracy might not be the best, but it does a surprisingly good job of detecting whether a given text was written by ChatGPT.


GPTRadar is another tool for checking AI works. Instead of developing a whole new algorithm or detection method, it uses the GPT framework as its detection method. This allows for great accuracy, while also giving metrics like probability and distribution.

GPTRadar has a fairly generous token-based model for running scans. Upon signing up, you’re awarded 2,000 free tokens. Additional tokens can be purchased for around $0.02, which means you can pay as you go.


Copyscape is similar in scope to Turnitin and sees a fair bit of use in academia and professional circuits alike. It can scan whole web pages, which is useful for detecting AI-generated works on the likes of blog posts.

Copyspace also allows for the scanning of Word documents and PDF files, meaning you’ve got a huge degree of flexibility. As an added bonus, you can check for actual plagiarism in addition to AI-powered content. If Turnitin isn’t cutting it for you, then Copyscape might be the way to go.


GPTZero is built expressly for detecting AI-written text. Compared to some AI detection tools, this one works fairly easily. You feed it a text and it uses statistical modeling to determine whether a piece was written by a person or an AI.

It has two key metrics that it gives per given piece, perplexity and burstiness. Perplexity is an analysis of the randomness of the text itself. Burstiness measures the actual variance found in the perplexity of a piece.

GPTZero can be prone to false positives but is fairly accurate. Do make sure to double-check with other tools if using it, however.

Closing Thoughts

So, can Turnitin detect ChatGPT? It does so with great accuracy, as a matter of fact. AI is likely here to stay, but the tools to keep writers in both academic and professional fields honest are likely to evolve alongside it.

It may be tempting to skip out on writing some lengthy essay using an AI like ChatGPT, but it isn’t a real reflection of your knowledge and voice. If you are considering doing so, keep in mind it is likely you’ll get caught at some point by an editor or instructor.

Summary Table

AI Detection ToolsDescription
TurnitinTurnitin can detect AI-generated content with a 98% accuracy rating. It can detect models like ChatGPT 3.5, ChatGPT 4, and Google’s Bard. It evaluates each sentence for AI generation and assigns a score between 0 and 1.
OpenAI Text ClassifierDeveloped by OpenAI, this tool can determine whether an AI or human wrote a given piece. It requires a minimum of 1,000 characters for analysis.
GPTRadarGPTRadar uses the GPT framework as its detection method. It provides metrics like probability and distribution. It operates on a token-based model for running scans.
CopyscapeCopyscape can scan whole web pages, Word documents, and PDF files. It can detect AI-generated works and check for actual plagiarism.
GPTZeroGPTZero uses statistical modeling to determine whether a piece was written by a person or an AI. It provides two key metrics: perplexity and burstiness.

Frequently Asked Questions

Can you use ChatGPT to write essays?

You can, given the proper prompt. However, you need to understand that ChatGPT has a finite source of knowledge to pull from, rather than doing your own research for a given topic.

Are AI detectors in use by colleges?

If your school uses Copyscape or Turnitin, it is absolutely using AI detection.

Can I use ChatGPT to generate content for my blog?

Sure, you could do that. There are certainly numerous tutorials on the web that emphasize how easy it is to write content. However, AI-generated content isn’t reflective of your own voice, experience, or knowledge.

Should I use ChatGPT for my professional writing?

No, especially if you’re working as a freelancer. Most clients hire writers with the understanding they have the expertise and knowledge to write about a given topic. If colleges are using AI detectors, you can guarantee your client will do so as well.

Is AI safe to use for written works?

That really does depend. For idle questions and queries, it can certainly work. There have been numerous stories as of late where malicious outputs resulted in serious consequences for the end user, however.

To top