Understand how our AI detection works and its limitations
Capybara uses OpenAI's language models to analyze text patterns and estimate whether content was generated by AI or written by a human.
When you submit text, our system:
Capybara provides probability estimates, not definitive proof. You should understand these limitations:
Human-written text may sometimes be flagged as AI-generated, especially if it's formal, technical, or follows common writing patterns.
AI-generated text that has been edited, paraphrased, or written with specific prompts may not be detected.
As AI writing tools improve, they become harder to detect. Our detection may lag behind the newest AI models.
Detection is less reliable for short passages (under 100 words). Longer text provides more patterns to analyze.
Text that combines human and AI writing will produce mixed results that may not reflect the actual ratio.
We believe in transparency. AI detection is an imperfect science, and we will always be honest about what our tool can and cannot do.
We provide probability scores rather than yes/no answers because the reality is nuanced. A 70% score means "likely AI-generated" not "definitely AI-generated."
Always use Capybara as one data point among many, not as the final word.
If you have questions about how Capybara works or concerns about a result, please reach out to us.
Email: support@saasita.space