GPTrue or False

GPTrue or False

Display the likelihood that a sample of text was generated by OpenAI's GPT-2 model.

  • GPTrue or False
  • GPTrue or False
What is GPTrue or False?
GPTrue or False is a Firefox add-on that utilizes OpenAI's GPT-2 model to detect the authenticity of text. Protect yourself from fake news and reviews by determining the likelihood that the selected text was generated by GPT-2.
Stats
Users: 47
Rating: 5.00 (2)
Version: 3.0.2 (Last updated: 2019-11-21)
Creation date: 2019-11-18
Weekly download count: NaN
Firefox on Android: No
Risk impact: Low risk impact
Risk likelihood: Low risk likelihood
Manifest version: 2
Permissions:
  • https://huggingface.co/*
  • activeTab
Size: 731.01K
URLs: Website
Stats date:

Other platforms

Not available on Chrome
Not available on Edge
Want to check extension ranking and stats more quickly for other Firefox add-ons? Install Chrome-Stats extension to view Chrome-Stats data as you browse the Firefox Browser Add-ons.
Chrome-Stats extension
Summary

OpenAI's recently released GPT-2 model has revealed itself to be capable of generating incredibly human-like text.

With the rampant spread of fake news and reviews in today's world, such tools may pose a threat to the quality of the information found on the internet.

Luckily, OpenAI also released a detector which is designed to detect whether a given portion of text has been generated by GPT-2 or not.

This extension brings the detector to your browser via a simple browser extension. Simply select a portion of text (at least 50 words) and the extension will determine and display the probability of whether the selected text was generated by GPT-2 or not.

Some notes:

  • The detector itself does not run in your browser. The extension makes HTTP requests to https://huggingface.co/openai-detector where the detector is publicly hosted and processes the response, displaying the results in your browser.
  • Results should be taken with a grain of salt. The detector is designed specifically to work with GPT-2 generated text. Using a different model or finetuning GPT-2 could be enough to avoid detection. In general, the detector is imperfect. Human-generated text may be marked as GPT-2 generated, and vice-versa.
    User reviews
    View all user reviews
    Safety
    Risk impact

    GPTrue or False is relatively safe to use as it requires very minimum permissions.

    Risk likelihood

    GPTrue or False has earned a fairly good reputation and likely can be trusted.

    Upgrade to see risk analysis details