Get started

AI Detector

Sapling's AI Detector estimates how likely text is to be AI-generated, both overall and sentence by sentence.

AI-generated text, also known as machine-generated text or automated text, is a product of natural language processing techniques that enable machines to mimic human language patterns and generate text. The benefits of AI-generated text include its ability to save time and resources by automating repetitive writing tasks and producing large volumes of content quickly. While AI-generated text can offer many benefits it is essential to be mindful of its limitations and potential risks when using it for specific purposes. Some issues that we've seen with current systems: Inaccurate / hallucinated facts Bland prose Biases that were present in the training data
Try a sample

Selectable text is required. Scanned images need OCR first.


Automate AI detection with the API

Try an API key, run an example, and review usage before choosing a paid plan.

Usage pricing and calculators · API billing is separate from Pro.

Get it everywhere you write

Use the AI detector extension to check text on other webpages. For grammar suggestions and writing assistance, install the Sapling writing extension.

Install AI Content Detector Install Sapling

Developed by former researchers at:
Cal Stanford
Meta Google

About Sapling's AI Detector

Sapling's free AI Detector estimates the probability that a text was written by a model such as ChatGPT, Claude, or Gemini. It gives an overall score, scores each sentence, and highlights the passages most likely to be AI-generated.

You can paste text directly, upload a PDF or Word file, or check a public web page, in English and 15 other languages. Free checks cover up to 5,000 characters without an account; Pro and Enterprise plans support up to 100,000. It's also available as a Chrome extension and through the AI Detector API.

Built for educators, SEO teams, and anyone screening user-generated content or AI slop.

How to use the free AI checker

Type or paste text above to score. Sapling's AI Detector becomes much more accurate after 50 or so words. The character count under the box updates as you type.

No current AI content detector (including Sapling's) should be used as a standalone check to determine whether text is AI-generated or written by a human. False positives and false negatives will occur.

  • The top section will show the overall score and highlight portions of the text that appear to be AI-generated.
  • The bottom section scores each sentence with the same detector model as the overall score, on the same scale: a higher score means the sentence is more likely AI-generated.
  • The overall score covers the whole text at once, so a text that is only partly AI-generated can have a low overall score while some of its sentences score high. Check both (along with your best judgment) to make an assessment.

Supported languages: English, Spanish, Portuguese, French, German, Italian, Dutch, Japanese, Polish, Chinese, Arabic, Russian, Swedish, Korean, Turkish and Romanian. The language is detected automatically; non-English text is scored by Sapling's multilingual model. Detector pages in other languages: Español · Português · Français · Deutsch · Italiano · Nederlands · 日本語 · Polski · 简体中文 · العربية · Русский · Svenska · 한국어 · Türkçe · Română.

Read the guide to reading and evaluating AI detector scores, how AI detectors work, and whether AI detectors are accurate.

Looking for other ways to score content? Contact us.

Get the AI detector whitepaper

PDF · 11 pages
First page of Sapling's AI Content Detection whitepaper Page 1 of 11 · PDF

Sapling AI Content Detection: Technical Overview and Validation Report documents how the detector works and how it performs.

  • Detection methodology: token-level classification, with document, sentence, and AI-fraction scores from the same model
  • Training data and model development
  • Evaluation results on a 1,000-sample held-out set, including false positive and false negative analysis
  • API, infrastructure, and data privacy details
  • Guidance on responsible use as a screening tool

Changelog

New

  • Improved English detection accuracy, especially for short fiction and magazine articles.
  • New English model that flags far less human-written text as AI in academic abstracts, personal essays and essays by English learners, and catches more AI-written text, including text copied from a PDF with line breaks inside paragraphs.
  • Updated the model for all 15 supported non-English languages. It flags less human-written text as AI, most of all in Dutch, Chinese and Russian, and catches more AI-written text.
  • Added AI detection for Polish, Chinese, Arabic, Russian, Swedish, Korean, Turkish and Romanian, and updated the model for Spanish, Portuguese, French, German, Italian, Dutch and Japanese.
  • More accurate models in every supported language, which flag far less human-written text with bullets or headings as AI. The English model also catches more text from recent models such as Claude 5.5. Italian and Dutch are now fully supported.
  • Added dedicated detectors for resumes and reviews. The review detector can check a batch of reviews or a CSV file at once.

Older updates

  • Added preliminary support for Italian and Dutch.
  • Added AI detection for Spanish, Portuguese, French, German and Japanese. The language is detected automatically and the text is scored by a model trained on human-written and AI-generated text in that language.
  • Higher precision on recent models and significantly higher detection of humanized AI text. On internal held-out evaluations, the false-positive rate fell by 60% relative to the previous model. Updated models including ChatGPT, Claude, Gemini, DeepSeek, and Qwen, and Kimi coverage was added.
  • Added sample texts — try detecting AI-generated generic text, AI-generated resumes, or human-written content directly from the tool.
  • Enhanced support for latest AI models including GPT-5, Claude 4.5, Gemini 2.5, Qwen3, and DeepSeek-V3. Maintained compatibility with older models.
  • Increased context length to 100,000 characters.

In development

  • Fewer false positives on college application essays and short essays by English learners.
  • Support for more languages.



Frequently Asked Questions

What is an AI content detector?

An AI content detector screens writing for patterns associated with machine-generated text and helps identify content that may have been written by an AI model. Sapling's AI Detector does this for pasted text, PDF and Word files, and web pages.

Advanced models such as OpenAI's GPT-6, Anthropic's Claude 5.5, Google's Gemini 3, DeepSeek-V3.2, Kimi K2.5, Qwen3, and other state-of-the-art AI systems have led to a rise in sophisticated machine-generated content, sometimes called "AI slop." This synthetic content is increasingly hard to tell apart from human writing, which raises a common question: "Was this written by AI?"

Despite rapid progress, these models still hallucinate facts, and they make it easier to cheat in writing and language courses. Low-quality AI-generated content, or slop, can be particularly problematic when it floods search results, social media, and other platforms.

This AI checker and slop detector tool provides a way of screening whether a piece of content is written by a human or a machine, helping resolve the "Is this AI-generated?" question and identify AI slop.

Is this AI checker free?

Yes. Sapling's AI detector is free to use without an account, for up to 5,000 characters per check. A free account adds higher rate limits and a history of your checks. Pro and Enterprise plans check up to 100,000 characters at a time.

How does this detector work?

Sapling's detector uses a Transformer model to estimate the probability that each token and sentence in the input is AI-generated. It uses a similar architecture to the models that generate AI text, but instead of predicting the next word, it scores each token. Sentence scores combine those token probabilities, so the overall result and per-sentence highlights stay on the same scale. Read more about how AI detectors work.

Do you have an AI detector for PDFs? Or an AI detector for Word .DOCX files?

Yes. Upload a PDF or DOCX file and Sapling extracts its selectable text before running AI detection. Use the file upload option under the editor, or try the PDF & Word AI Detector to inspect the extracted text next to the score.

Can I use the AI detector on resumes and job applications?

Yes! Paste resume or cover-letter text directly into the editor above, or click Try a sample → AI-generated resume to see how the detector performs on a typical AI-written resume.

For high-volume screening workflows, our AI Detector API makes it easy to integrate resume detection into your applicant-tracking system. Visit our AI Resume Detector page to learn more or contact us about enterprise plans. Checking product or app reviews instead? Use the AI Review Detector. Checking student essays? Use the AI Detector for Teachers.

Is there an AI Content Detector extension I can use on other websites?

Yes! You can install the AI Content Detector for ChatGPT extension by Sapling.ai. This extension will allow you to check for AI-generated content anywhere on the web. Select text on any webpage, then click the Detect AI button to see a complete analysis of the selected text.

When using chatbots such as ChatGPT, the Detect AI button will be embedded next to each generated result allowing you to easily run AI detection with a single click.

You will also be able to edit the analyzed text and recheck your work. This will allow you to easily fix the sections that have been flagged as AI-generated.

How accurate is the AI detector?

Accuracy varies with the dataset, text length, and decision threshold. Measure false positives and missed detections on examples from your workflow.

See the evaluation guide and technical whitepaper for methodology and reported results. Check the model version and dataset before applying a benchmark to a current workflow. For background, read Are AI detectors accurate?

I pasted my own writing but it says it's AI-generated! What's going on?

Sapling's detector can have false positives. The shorter the text is, the more general it is, and the more essay-like it is, the more likely it is to result in a false positive. We are working on improving the system so that this occurs less frequently.

My text is getting cut off. How can I run the AI detector on longer texts?

Sapling's free AI Detector checks up to 5,000 characters (roughly 1,000 to 1,250 tokens). Pro and Enterprise Sapling subscribers can paste texts of up to 100,000 characters (roughly 20,000 to 25,000 tokens). For texts longer than that, please break up the text into multiple sections, or consider using our API. If you plan to process more than 5 million characters/month, contact us to see how we can better support your use case.

Can I access the AI detector through an API?

Yes. Sapling's AI Detector is available through an API. See the AI Detector API documentation.

Does the AI detector work with languages other than English?

Yes, for Spanish, Portuguese, French, German, Italian, Dutch, Japanese, Polish, Chinese, Arabic, Russian, Swedish, Korean, Turkish and Romanian. Sapling detects the language of the text automatically and scores those languages with its multilingual model, trained on human-written and AI-generated text. Text in other languages is scored by the English model and is less reliable. The same applies to the AI Detector API.

Does the AI check work with outputs from different language models such as GPT-6, Claude 5.5, Gemini 3, DeepSeek-V3.2, Kimi K2.5, Qwen3, and other AI models?

While language models are evolving and have their differences, they usually use a similar machine learning architecture and a similar dataset on which they're trained. Hence, even AI detectors trained on different and earlier versions of language model outputs should perform significantly better than random on other models.

That said, to get the best performance, detectors should be trained on the outputs of the latest systems. Sapling regularly retrains and finetunes its detector to keep it up-to-date with cutting-edge AI models including GPT-6, Claude 5.5, Gemini 3, DeepSeek-V3.2, Kimi K2.5, Qwen3, and other emerging systems, while maintaining compatibility with older models.

How does Sapling's AI detector compare to others such as Pangram, Copyleaks, GPTZero, Originality.ai, or Turnitin?

Our AI detection API comparison lines up Sapling, Pangram, GPTZero, Copyleaks, Originality.ai, Grammarly, Winston AI and Turnitin on features and price. We also have side-by-side pages for GPTZero and Pangram, and a comparison of the best AI detectors by use case.

Detectors are trained on different data, so the best test is still your own: collect a small dataset of a dozen or so examples (such as blog posts and essays) and try for yourself :-).

What is AI slop and how does the slop detector work?

"AI slop" refers to low-quality, generic, and/or repetitive content mass-produced by AI language models. It often appears in sites with user-generated content such as product reviews, social media posts, and articles. Slop is characterized by bland phrasing, filler text, and a lack of genuine insight or originality.

Do tools that claim to make AI-generated text undetectable work?

We've seen such tools make false claims such as that a text "passed" Sapling.ai's detector even though no check was performed by Sapling.ai. Please be careful when using such tools.