Home > Apps

AI Text Classifier – OpenAI’s release explained

Everything we know about OpenAI's new AI detector
Last Updated on February 2, 2023
PC Guide is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Read More
You can trust PC Guide: Our team of experts use a combination of independent consumer research, in-depth testing where appropriate - which will be flagged as such, and market analysis when recommending products, software and services. Find out how we test here.

Big news! OpenAI launches a new tool that can detect AI-written text. Their AI Text Classifier became live on the 31st of January and is free to use for those looking to determine whether a chunk of text was AI-generated or not.

At the moment, only the beta version is available. OpenAI is still developing the tool, so they’ve released it to get some extra feedback from their users.

Could this be the tool educators have been hoping for? Well, according to OpenAI the classifier isn’t perfect and we have outlined its limitations below. Keep reading to learn about the limitations and how to use the tool.

AI Text Classifier limitations

There are a few problems with OpenAI’s AI Text Classifier that you need to be aware of. The first and most important one is that the classifier is not ‘fully reliable’.

From what it seems, the tool isn’t too accurate. According to OpenAI, during testing, the classifier only correctly identified 26% of AI-generated texts. Moreover, the tool made mistakes in identifying 9% of human-written text as being AI-generated.

OpenAI also lists the other limitations of the tool on its website. Here is a summary of them below:

  • The tool requires an input of at least 1,000 characters as the classifier is unreliable on short texts.
  • The classifier works best on English text and works poorly on other languages.
  • Predictable text such as numbers in a sequence is impossible to classify.
  • AI language models can be altered to become undetectable by AI classifiers, which raises concerns about the long-term effectiveness of OpenAI’s tool.
  • If the inputted text is very different from the training dataset, the classifier may produce the wrong result.

How to use OpenAI’s AI Text Classifier

You can use OpenAI’s AI Text Classifier for free on their website. Despite the classifier not being entirely reliable, the AI company has made it publicly available to gain feedback on whether this tool would be helpful.

Using the classifier is very simple. Here are a few steps to get you started:

  1. Head to the OpenAI AI Text Classifier website.
  2. Create an account or sign in.
  3. Copy and paste the text you would like to classify into the ‘Text’ box and click ‘Submit’.

After you have submitted the text, OpenAI will output an AI-generated likelihood score. Scores come in the form of ‘very unlikely’, ‘unlikely’, ‘unclear if it is’, ‘possibly’, and ‘likely’.

Keep in mind, that the tool is not 100% accurate so there is a chance you may get a false positive or false negative result.

Final Thoughts

Since the release of Chat GPT last year, it was only a matter of time before OpenAI came out with its version of an AI detector.

The OpenAI AI Text Classifier is currently a work in progress but may be useful to try out. If you are looking for other AI text detectors, why not read our Chat GPT detectors article here?

Funmi joined PC Guide in November 2022, and was a driving force for the site's ChatGPT coverage. She has a wide knowledge of AI apps, gaming and consumer technology.